In [1]:

    

%pylab inline


    

    




Populating the interactive namespace from numpy and matplotlib

In [2]:

    

import sys
sys.path.insert(0, "../")


    

In [3]:

    

from collections import OrderedDict
import numpy
import pandas

from sklearn.ensemble import RandomForestClassifier
from sklearn.metrics import roc_curve, roc_auc_score

from rep.metaml import FoldingClassifier


    

    




/mnt/mfs/miniconda/envs/rep_py2/lib/python2.7/site-packages/sklearn/cross_validation.py:44: DeprecationWarning: This module was deprecated in version 0.18 in favor of the model_selection module into which all the refactored classes and functions are moved. Also note that the interface of the new CV iterators are different from that of this module. This module will be removed in 0.20.
  "This module will be removed in 0.20.", DeprecationWarning)

In [4]:

    

from utils import get_N_B_events, predict_by_estimator, bootstrap_calibrate_prob
from utils import get_events_number, get_events_statistics, result_table, compute_mistag


    

In [5]:

    

import root_numpy
def read_samples(filename, filename_vtx):
    datasets = dict()
    for name, selection, f in zip(['K', 'e', 'mu', 'vtx'], 
                                  ['K_cut == 1', 'e_cut == 1', 'mu_cut == 1', '(v_cut == 1) & (vcharge > 0.2)'],
                                  [filename] * 3 + [filename_vtx]):
        data = pandas.DataFrame(root_numpy.root2array(f, selection=selection))                       
        if name == 'vtx':
            data['label'] = (data.signB.values * data.signVtx.values < 0) * 1
        else:
            data['label'] = (data.signB.values * data.signTrack.values < 0) * 1
        data['event_id'] = data.run.apply(str) + '_' + data.event.apply(int).apply(str)
        data['tagAnswer'] = data.signB * (2 * data.label - 1)
        data['N_sig_sw'] = 1.
        datasets[name] = data
    
    return datasets


    

In [6]:

    

def compute_efficiencies(datasets, N_B_events):
    result = dict()
    for key, data in datasets.items():
        N_B_passed = get_events_number(data)
        result[key] = (1. * N_B_passed / N_B_events, sqrt(N_B_passed) / N_B_events)
    return result


    

In [ ]:

    

datasets = read_samples('../datasets/MC/csv/WG/Bu_JPsiK/2012/Tracks.root',
                        '../datasets/MC/csv/WG/Bu_JPsiK/2012/Vertices.root')


    

In [ ]:

    

datasets_kstar = read_samples('../datasets/MC/csv/WG/Bd_JPsiKstar/2012/Tracks.root',
                              '../datasets/MC/csv/WG/Bd_JPsiKstar/2012/Vertices.root')
datasets_ks = read_samples('../datasets/MC/csv/WG/Bd_JPsiKs/2012/Tracks.root',
                           '../datasets/MC/csv/WG/Bd_JPsiKs/2012/Vertices.root')


    

In [ ]:

    

from utils import compute_N_B_events_MC
N_B_events = compute_N_B_events_MC('../datasets/MC/csv/WG/Bu_JPsiK/2012/Tracks.root',
                                   '../datasets/MC/csv/WG/Bu_JPsiK/2012/Vertices.root')
N_B_events_kstar = compute_N_B_events_MC('../datasets/MC/csv/WG/Bd_JPsiKstar/2012/Tracks.root',
                                         '../datasets/MC/csv/WG/Bd_JPsiKstar/2012/Vertices.root')
N_B_events_ks = compute_N_B_events_MC('../datasets/MC/csv/WG/Bd_JPsiKs/2012/Tracks.root',
                                      '../datasets/MC/csv/WG/Bd_JPsiKs/2012/Vertices.root')


    

In [ ]:

    

N_B_events, N_B_events_kstar, N_B_events_ks


    

In [ ]:

    

eff = compute_efficiencies(datasets, N_B_events)
eff_kstar = compute_efficiencies(datasets_kstar, N_B_events_kstar)
eff_ks = compute_efficiencies(datasets_ks, N_B_events_ks)


    

In [ ]:

    

print 'K+-'
print pandas.DataFrame(eff)
print 'K*'
print pandas.DataFrame(eff_kstar)
print 'Ks'
print pandas.DataFrame(eff_ks)


    

In [ ]:

    

features_vtx = ['mult', 'nnkrec', 'log_ptB: log(ptB)', 'vflag', 'log_ipsmean: log(ipsmean)', 
                'log_ptmean: log(ptmean)', 'vcharge', 
                'log_svm: log(svm)', 'log_svp: log(svp)', 'BDphiDir', 'log_svtau: log(svtau)', 'docamax']

features_ele = ['mult', 'log_partPt: log(partPt)', 'log_partP: log(partP)',
                'log_ptB: log(ptB)', 'log_IPs: log(IPs)', 'partlcs', 'log_eOverP: log(EOverP)', 
                'ghostProb', 'log_IPPU: log(IPPU)']
features_muon = ['mult', 'log_partPt: log(partPt)', 'log_partP: log(partP)',
                'log_ptB: log(ptB)', 'log_IPs: log(IPs)', 'partlcs', 'PIDNNm', 'ghostProb', 'log_IPPU: log(IPPU)']
features_kaon = ['mult', 'log_partPt: log(partPt)', 'log_partP: log(partP)',
                 'nnkrec','log_ptB: log(ptB)', 'log_IPs: log(IPs)', 'partlcs', 
                'PIDNNk', 'PIDNNpi', 'PIDNNp', 'ghostProb', 'log_IPPU: log(IPPU)']
                                    
features = {'e': features_ele, 'mu': features_muon,
            'K': features_kaon, 'vtx': features_vtx}

# features_tr = ['mult', 'log_partPt: log(partPt)', 'log_partP: log(partP)',
#                'log_ptB: log(ptB)', 'log_IPs: log(IPs)', 'partlcs', 'ghostProb', 'log_IPPU: log(IPPU)', 
#                'log_EOverP: log(EOverP)', 
#                'PIDNNpi', 'PIDNNp', 'PIDNNk', 'PIDNNm', 'PIDNNe', 'nnkrec', 'log_partTheta: log(partTheta)',
#                'log_partPhi: log(partPhi)', 'veloch']


    

In [ ]:

    

estimators = OrderedDict()


    

In [15]:

    

from rep.estimators import XGBoostClassifier
from rep.metaml import FoldingClassifier

xgb_base_ele = XGBoostClassifier(colsample=0.8, eta=0.01, nthreads=4, 
                                 n_estimators=200, subsample=0.5, max_depth=5) 

xgb_base_other = XGBoostClassifier(colsample=0.8, eta=0.01, nthreads=4, 
                                   n_estimators=500, subsample=0.5, max_depth=5) 

for key, data in datasets.items():
    if 'e' in key:
        xgb_base = xgb_base_ele
    else:
        xgb_base = xgb_base_other
        
    estimators[key + '_xgboost'] = FoldingClassifier(xgb_base, n_folds=2, random_state=523,
                                                     features=features[key])
    estimators[key + '_xgboost'].fit(data, data['label'], data['N_sig_sw'])


    

In [16]:

    

from rep.estimators import TMVAClassifier
from rep.metaml import FoldingClassifier

tmva_base_muon = TMVAClassifier(method='kMLP', factory_options='Transformations=I,D,N', sigmoid_function='identity',
                                NeuronType='tanh', NCycles=280, HiddenLayers='N+5', TrainingMethod='BFGS', TestRate=5,
                                UseRegulator=True, EstimatorType='CE')

tmva_base_ele = TMVAClassifier(method='kMLP', factory_options='Transformations=I,D,N', sigmoid_function='identity',
                               NeuronType='sigmoid', NCycles=180, HiddenLayers='N+5', TrainingMethod='BFGS', 
                               UseRegulator=True)

tmva_base_kaon_vtx = TMVAClassifier(method='kMLP', factory_options='Transformations=I,D,N', 
                                    sigmoid_function='identity',
                                    NeuronType='tanh', NCycles=180, HiddenLayers='N+5', TrainingMethod='BFGS', 
                                    UseRegulator=True, EstimatorType='CE')


for key, data in datasets.items():
    if 'e' in key:
        tmva_base = tmva_base_ele
    elif 'mu' in key:
        tmva_base = tmva_base_muon
    else:
        tmva_base = tmva_base_kaon_vtx
    estimators[key + '_tmva'] = FoldingClassifier(tmva_base, n_folds=2, random_state=523, features=features[key])
    estimators[key + '_tmva'].fit(data, data['label'], data['N_sig_sw'])


    

    




#error "You need a ISO C conforming compiler to use the glibc headers"
*** Interpreter error recovered ***
--- Factory                  : You are running ROOT Version: 5.34/32, Jun 23, 2015
--- Factory                  : 
--- Factory                  : _/_/_/_/_/ _|      _|  _|      _|    _|_|   
--- Factory                  :    _/      _|_|  _|_|  _|      _|  _|    _| 
--- Factory                  :   _/       _|  _|  _|  _|      _|  _|_|_|_| 
--- Factory                  :  _/        _|      _|    _|  _|    _|    _| 
--- Factory                  : _/         _|      _|      _|      _|    _| 
--- Factory                  : 
--- Factory                  : ___________TMVA Version 4.2.0, Sep 19, 2013
--- Factory                  : 
--- DataSetInfo              : Added class "Signal"	 with internal class number 0
--- DataSetInfo              : Added class "Background"	 with internal class number 1
--- Factory                  : Add Tree TrainAssignTree_Signal of type Signal with 25886 events
--- Factory                  : Add Tree TestAssignTree_Signal of type Signal with 25886 events
--- Factory                  : Add Tree TrainAssignTree_Background of type Background with 11737 events
--- Factory                  : Add Tree TestAssignTree_Background of type Background with 11737 events
--- DataSetInfo              : Class index : 0  name : Signal
--- DataSetInfo              : Class index : 1  name : Background
--- Factory                  : Booking method: REP_Estimator
--- REP_Estimator            : Building Network
--- REP_Estimator            : Initializing weights
--- DataSetFactory           : Splitmode is: "RANDOM" the mixmode is: "SAMEASSPLITMODE"
--- DataSetFactory           : Create training and testing trees -- looping over class "Signal" ...
--- DataSetFactory           : Weight expression for class 'Signal': "weight"
--- DataSetFactory           : Create training and testing trees -- looping over class "Background" ...
--- DataSetFactory           : Weight expression for class 'Background': "weight"
--- DataSetFactory           : Number of events in input trees (after possible flattening of arrays):
--- DataSetFactory           :     Signal          -- number of events       : 51772  / sum of weights: 51772
--- DataSetFactory           :     Background      -- number of events       : 23474  / sum of weights: 23474
--- DataSetFactory           :     Signal     tree -- total number of entries: 51772
--- DataSetFactory           :     Background tree -- total number of entries: 23474
--- DataSetFactory           : Preselection: (will NOT affect number of requested training and testing events)
--- DataSetFactory           :     Signal     requirement: "1"
--- DataSetFactory           :     Signal          -- number of events passed: 51772  / sum of weights: 51772
--- DataSetFactory           :     Signal          -- efficiency             : 1     
--- DataSetFactory           :     Background requirement: "1"
--- DataSetFactory           :     Background      -- number of events passed: 23474  / sum of weights: 23474
--- DataSetFactory           :     Background      -- efficiency             : 1     
--- DataSetFactory           : Weight renormalisation mode: "EqualNumEvents": renormalises all event classes ...
--- DataSetFactory           :  such that the effective (weighted) number of events in each class is the same 
--- DataSetFactory           :  (and equals the number of events (entries) given for class=0 )
--- DataSetFactory           : ... i.e. such that Sum[i=1..N_j]{w_i} = N_classA, j=classA, classB, ...
--- DataSetFactory           : ... (note that N_j is the sum of TRAINING events
--- DataSetFactory           :  ..... Testing events are not renormalised nor included in the renormalisation factor!)
--- DataSetFactory           : --> Rescale Signal     event weights by factor: 1
--- DataSetFactory           : --> Rescale Background event weights by factor: 2.2055
--- DataSetFactory           : Number of training and testing events after rescaling:
--- DataSetFactory           : ------------------------------------------------------
--- DataSetFactory           : Signal     -- training events            : 25886 (sum of weights: 25886) - requested were 0 events
--- DataSetFactory           : Signal     -- testing events             : 25886 (sum of weights: 25886) - requested were 0 events
--- DataSetFactory           : Signal     -- training and testing events: 51772 (sum of weights: 51772)
--- DataSetFactory           : Background -- training events            : 11737 (sum of weights: 25886) - requested were 0 events
--- DataSetFactory           : Background -- testing events             : 11737 (sum of weights: 11737) - requested were 0 events
--- DataSetFactory           : Background -- training and testing events: 23474 (sum of weights: 37623)
--- DataSetFactory           : Create internal training tree
--- DataSetFactory           : Create internal testing tree
--- DataSetInfo              : Correlation matrix (Signal):
--- DataSetInfo              : -------------------------------------------------------------------------------------------
--- DataSetInfo              :                mult log_partPt log_partP log_ptB log_IPs partlcs  PIDNNm ghostProb log_IPPU
--- DataSetInfo              :       mult:  +1.000     +0.008    +0.017  +0.070  +0.029  +0.113  -0.121    +0.108   -0.015
--- DataSetInfo              : log_partPt:  +0.008     +1.000    +0.493  +0.042  +0.001  -0.050  +0.362    -0.067   +0.063
--- DataSetInfo              :  log_partP:  +0.017     +0.493    +1.000  +0.014  -0.067  -0.043  +0.179    +0.007   -0.144
--- DataSetInfo              :    log_ptB:  +0.070     +0.042    +0.014  +1.000  -0.005  +0.026  -0.012    +0.024   +0.001
--- DataSetInfo              :    log_IPs:  +0.029     +0.001    -0.067  -0.005  +1.000  -0.067  +0.135    -0.080   -0.036
--- DataSetInfo              :    partlcs:  +0.113     -0.050    -0.043  +0.026  -0.067  +1.000  -0.397    +0.635   -0.061
--- DataSetInfo              :     PIDNNm:  -0.121     +0.362    +0.179  -0.012  +0.135  -0.397  +1.000    -0.428   +0.087
--- DataSetInfo              :  ghostProb:  +0.108     -0.067    +0.007  +0.024  -0.080  +0.635  -0.428    +1.000   -0.078
--- DataSetInfo              :   log_IPPU:  -0.015     +0.063    -0.144  +0.001  -0.036  -0.061  +0.087    -0.078   +1.000
--- DataSetInfo              : -------------------------------------------------------------------------------------------
--- DataSetInfo              : Correlation matrix (Background):
--- DataSetInfo              : -------------------------------------------------------------------------------------------
--- DataSetInfo              :                mult log_partPt log_partP log_ptB log_IPs partlcs  PIDNNm ghostProb log_IPPU
--- DataSetInfo              :       mult:  +1.000     +0.020    +0.009  +0.081  -0.060  +0.108  -0.161    +0.094   -0.005
--- DataSetInfo              : log_partPt:  +0.020     +1.000    +0.475  +0.066  +0.070  -0.024  +0.284    -0.016   +0.056
--- DataSetInfo              :  log_partP:  +0.009     +0.475    +1.000  +0.039  +0.004  -0.046  +0.133    +0.029   -0.153
--- DataSetInfo              :    log_ptB:  +0.081     +0.066    +0.039  +1.000  -0.011  +0.029  -0.010    +0.015   +0.005
--- DataSetInfo              :    log_IPs:  -0.060     +0.070    +0.004  -0.011  +1.000  -0.115  +0.238    -0.116   -0.071
--- DataSetInfo              :    partlcs:  +0.108     -0.024    -0.046  +0.029  -0.115  +1.000  -0.419    +0.661   -0.047
--- DataSetInfo              :     PIDNNm:  -0.161     +0.284    +0.133  -0.010  +0.238  -0.419  +1.000    -0.429   +0.090
--- DataSetInfo              :  ghostProb:  +0.094     -0.016    +0.029  +0.015  -0.116  +0.661  -0.429    +1.000   -0.064
--- DataSetInfo              :   log_IPPU:  -0.005     +0.056    -0.153  +0.005  -0.071  -0.047  +0.090    -0.064   +1.000
--- DataSetInfo              : -------------------------------------------------------------------------------------------
--- DataSetFactory           :  
--- Factory                  : 
--- Factory                  : current transformation string: 'I,D,N'
--- Factory                  : Create Transformation "I" with events from all classes.
--- Id                       : Transformation, Variable selection : 
--- Id                       : Input : variable 'mult' (index=0).   <---> Output : variable 'mult' (index=0).
--- Id                       : Input : variable 'log_partPt' (index=1).   <---> Output : variable 'log_partPt' (index=1).
--- Id                       : Input : variable 'log_partP' (index=2).   <---> Output : variable 'log_partP' (index=2).
--- Id                       : Input : variable 'log_ptB' (index=3).   <---> Output : variable 'log_ptB' (index=3).
--- Id                       : Input : variable 'log_IPs' (index=4).   <---> Output : variable 'log_IPs' (index=4).
--- Id                       : Input : variable 'partlcs' (index=5).   <---> Output : variable 'partlcs' (index=5).
--- Id                       : Input : variable 'PIDNNm' (index=6).   <---> Output : variable 'PIDNNm' (index=6).
--- Id                       : Input : variable 'ghostProb' (index=7).   <---> Output : variable 'ghostProb' (index=7).
--- Id                       : Input : variable 'log_IPPU' (index=8).   <---> Output : variable 'log_IPPU' (index=8).
--- Factory                  : Create Transformation "D" with events from all classes.
--- Deco                     : Transformation, Variable selection : 
--- Deco                     : Input : variable 'mult' (index=0).   <---> Output : variable 'mult' (index=0).
--- Deco                     : Input : variable 'log_partPt' (index=1).   <---> Output : variable 'log_partPt' (index=1).
--- Deco                     : Input : variable 'log_partP' (index=2).   <---> Output : variable 'log_partP' (index=2).
--- Deco                     : Input : variable 'log_ptB' (index=3).   <---> Output : variable 'log_ptB' (index=3).
--- Deco                     : Input : variable 'log_IPs' (index=4).   <---> Output : variable 'log_IPs' (index=4).
--- Deco                     : Input : variable 'partlcs' (index=5).   <---> Output : variable 'partlcs' (index=5).
--- Deco                     : Input : variable 'PIDNNm' (index=6).   <---> Output : variable 'PIDNNm' (index=6).
--- Deco                     : Input : variable 'ghostProb' (index=7).   <---> Output : variable 'ghostProb' (index=7).
--- Deco                     : Input : variable 'log_IPPU' (index=8).   <---> Output : variable 'log_IPPU' (index=8).
--- Factory                  : Create Transformation "N" with events from all classes.
--- Norm                     : Transformation, Variable selection : 
--- Norm                     : Input : variable 'mult' (index=0).   <---> Output : variable 'mult' (index=0).
--- Norm                     : Input : variable 'log_partPt' (index=1).   <---> Output : variable 'log_partPt' (index=1).
--- Norm                     : Input : variable 'log_partP' (index=2).   <---> Output : variable 'log_partP' (index=2).
--- Norm                     : Input : variable 'log_ptB' (index=3).   <---> Output : variable 'log_ptB' (index=3).
--- Norm                     : Input : variable 'log_IPs' (index=4).   <---> Output : variable 'log_IPs' (index=4).
--- Norm                     : Input : variable 'partlcs' (index=5).   <---> Output : variable 'partlcs' (index=5).
--- Norm                     : Input : variable 'PIDNNm' (index=6).   <---> Output : variable 'PIDNNm' (index=6).
--- Norm                     : Input : variable 'ghostProb' (index=7).   <---> Output : variable 'ghostProb' (index=7).
--- Norm                     : Input : variable 'log_IPPU' (index=8).   <---> Output : variable 'log_IPPU' (index=8).
--- Id                       : Preparing the Identity transformation...
--- Deco                     : Preparing the Decorrelation transformation...
--- Norm                     : Preparing the transformation.
--- TFHandler_Factory        : ---------------------------------------------------------------------
--- TFHandler_Factory        :   Variable          Mean          RMS   [        Min          Max ]
--- TFHandler_Factory        : ---------------------------------------------------------------------
--- TFHandler_Factory        :       mult:    -0.55354     0.21476   [     -1.0000      1.0000 ]
--- TFHandler_Factory        : log_partPt:    -0.23970     0.31827   [     -1.0000      1.0000 ]
--- TFHandler_Factory        :  log_partP:   -0.067719     0.35265   [     -1.0000      1.0000 ]
--- TFHandler_Factory        :    log_ptB:     0.28756     0.21612   [     -1.0000      1.0000 ]
--- TFHandler_Factory        :    log_IPs:     0.12723     0.23528   [     -1.0000      1.0000 ]
--- TFHandler_Factory        :    partlcs:    -0.20213     0.20918   [     -1.0000      1.0000 ]
--- TFHandler_Factory        :     PIDNNm:     0.21292     0.28937   [     -1.0000      1.0000 ]
--- TFHandler_Factory        :  ghostProb:    -0.70228     0.10744   [     -1.0000      1.0000 ]
--- TFHandler_Factory        :   log_IPPU:     0.11268     0.52816   [     -1.0000      1.0000 ]
--- TFHandler_Factory        : ---------------------------------------------------------------------
--- TFHandler_Factory        : Plot event variables for Id_Deco_Norm
--- TFHandler_Factory        : Create scatter and profile plots in target-file directory: 
--- TFHandler_Factory        : /mnt/mfs/notebook/analyses/tagging_LHCb/MC/tmpsjMIuu/result.root:/InputVariables_Id_Deco_Norm/CorrelationPlots
--- TFHandler_Factory        :  
--- TFHandler_Factory        : Ranking input variables (method unspecific)...
--- Id_Deco_NormTransforma...: Ranking result (top variable is best ranked)
--- Id_Deco_NormTransforma...: -----------------------------------
--- Id_Deco_NormTransforma...: Rank : Variable   : Separation
--- Id_Deco_NormTransforma...: -----------------------------------
--- Id_Deco_NormTransforma...:    1 : log_partPt : 1.171e-02
--- Id_Deco_NormTransforma...:    2 : PIDNNm     : 9.809e-03
--- Id_Deco_NormTransforma...:    3 : log_IPs    : 7.189e-03
--- Id_Deco_NormTransforma...:    4 : log_partP  : 5.671e-03
--- Id_Deco_NormTransforma...:    5 : mult       : 5.630e-03
--- Id_Deco_NormTransforma...:    6 : ghostProb  : 4.175e-03
--- Id_Deco_NormTransforma...:    7 : partlcs    : 3.174e-03
--- Id_Deco_NormTransforma...:    8 : log_IPPU   : 2.220e-03
--- Id_Deco_NormTransforma...:    9 : log_ptB    : 2.009e-03
--- Id_Deco_NormTransforma...: -----------------------------------
--- Factory                  :  
--- Factory                  : Train all methods for Classification ...
--- Factory                  : Train method: REP_Estimator for Classification
--- REP_Estimator            : Begin training
--- REP_Estimator            : Training Network
Error in <TDecompLU::InvertLU>: matrix is singular, 8 diag elements < tolerance of 2.2204e-16
Error in <TDecompLU::InvertLU>: matrix is singular, 1 diag elements < tolerance of 2.2204e-16
Error in <TDecompLU::InvertLU>: matrix is singular, 1 diag elements < tolerance of 2.2204e-16
Error in <TDecompLU::InvertLU>: matrix is singular, 2 diag elements < tolerance of 2.2204e-16
--- REP_Estimator            : Finalizing handling of Regulator terms, trainE=1.35513 testE=1.35552
--- REP_Estimator            : Done with handling of Regulator terms
--- REP_Estimator            : End of training                                              
--- REP_Estimator            : Elapsed time for training with 37623 events: 387 sec         
--- REP_Estimator            : Create MVA output for classification on training sample
--- REP_Estimator            : Evaluation of REP_Estimator on training sample (37623 events)
--- REP_Estimator            : Elapsed time for evaluation of 37623 events: 0.137 sec       
--- REP_Estimator            : Creating weight file in xml format: weights/TMVAEstimation_REP_Estimator.weights.xml
--- REP_Estimator            : Creating standalone response class: weights/TMVAEstimation_REP_Estimator.class.C
--- REP_Estimator            : Write special histos to file: /mnt/mfs/notebook/analyses/tagging_LHCb/MC/tmpsjMIuu/result.root:/Method_MLP/REP_Estimator
--- Factory                  : Training finished
--- Factory                  : 
--- Factory                  : Ranking input variables (method specific)...
--- REP_Estimator            : Ranking result (top variable is best ranked)
--- REP_Estimator            : -----------------------------------
--- REP_Estimator            : Rank : Variable   : Importance
--- REP_Estimator            : -----------------------------------
--- REP_Estimator            :    1 : mult       : 2.933e+04
--- REP_Estimator            :    2 : log_IPPU   : 4.861e+02
--- REP_Estimator            :    3 : log_partP  : 1.126e+02
--- REP_Estimator            :    4 : log_IPs    : 7.576e+01
--- REP_Estimator            :    5 : log_ptB    : 4.506e+01
--- REP_Estimator            :    6 : partlcs    : 1.936e+01
--- REP_Estimator            :    7 : PIDNNm     : 1.034e+01
--- REP_Estimator            :    8 : log_partPt : 5.141e+00
--- REP_Estimator            :    9 : ghostProb  : 2.299e-02
--- REP_Estimator            : -----------------------------------
--- Factory                  : 
--- Factory                  : === Destroy and recreate all methods via weight files for testing ===
--- Factory                  : 
--- MethodBase               : Reading weight file: weights/TMVAEstimation_REP_Estimator.weights.xml
--- REP_Estimator            : Read method "REP_Estimator" of type "MLP"
--- REP_Estimator            : MVA method was trained with TMVA Version: 4.2.0
--- REP_Estimator            : MVA method was trained with ROOT Version: 5.34/32
--- REP_Estimator            : Building Network
--- REP_Estimator            : Initializing weights

#error "You need a ISO C conforming compiler to use the glibc headers"
*** Interpreter error recovered ***
--- Factory                  : You are running ROOT Version: 5.34/32, Jun 23, 2015
--- Factory                  : 
--- Factory                  : _/_/_/_/_/ _|      _|  _|      _|    _|_|   
--- Factory                  :    _/      _|_|  _|_|  _|      _|  _|    _| 
--- Factory                  :   _/       _|  _|  _|  _|      _|  _|_|_|_| 
--- Factory                  :  _/        _|      _|    _|  _|    _|    _| 
--- Factory                  : _/         _|      _|      _|      _|    _| 
--- Factory                  : 
--- Factory                  : ___________TMVA Version 4.2.0, Sep 19, 2013
--- Factory                  : 
--- DataSetInfo              : Added class "Signal"	 with internal class number 0
--- DataSetInfo              : Added class "Background"	 with internal class number 1
--- Factory                  : Add Tree TrainAssignTree_Signal of type Signal with 25757 events
--- Factory                  : Add Tree TestAssignTree_Signal of type Signal with 25757 events
--- Factory                  : Add Tree TrainAssignTree_Background of type Background with 11866 events
--- Factory                  : Add Tree TestAssignTree_Background of type Background with 11866 events
--- DataSetInfo              : Class index : 0  name : Signal
--- DataSetInfo              : Class index : 1  name : Background
--- Factory                  : Booking method: REP_Estimator
--- REP_Estimator            : Building Network
--- REP_Estimator            : Initializing weights
--- DataSetFactory           : Splitmode is: "RANDOM" the mixmode is: "SAMEASSPLITMODE"
--- DataSetFactory           : Create training and testing trees -- looping over class "Signal" ...
--- DataSetFactory           : Weight expression for class 'Signal': "weight"
--- DataSetFactory           : Create training and testing trees -- looping over class "Background" ...
--- DataSetFactory           : Weight expression for class 'Background': "weight"
--- DataSetFactory           : Number of events in input trees (after possible flattening of arrays):
--- DataSetFactory           :     Signal          -- number of events       : 51514  / sum of weights: 51514
--- DataSetFactory           :     Background      -- number of events       : 23732  / sum of weights: 23732
--- DataSetFactory           :     Signal     tree -- total number of entries: 51514
--- DataSetFactory           :     Background tree -- total number of entries: 23732
--- DataSetFactory           : Preselection: (will NOT affect number of requested training and testing events)
--- DataSetFactory           :     Signal     requirement: "1"
--- DataSetFactory           :     Signal          -- number of events passed: 51514  / sum of weights: 51514
--- DataSetFactory           :     Signal          -- efficiency             : 1     
--- DataSetFactory           :     Background requirement: "1"
--- DataSetFactory           :     Background      -- number of events passed: 23732  / sum of weights: 23732
--- DataSetFactory           :     Background      -- efficiency             : 1     
--- DataSetFactory           : Weight renormalisation mode: "EqualNumEvents": renormalises all event classes ...
--- DataSetFactory           :  such that the effective (weighted) number of events in each class is the same 
--- DataSetFactory           :  (and equals the number of events (entries) given for class=0 )
--- DataSetFactory           : ... i.e. such that Sum[i=1..N_j]{w_i} = N_classA, j=classA, classB, ...
--- DataSetFactory           : ... (note that N_j is the sum of TRAINING events
--- DataSetFactory           :  ..... Testing events are not renormalised nor included in the renormalisation factor!)
--- DataSetFactory           : --> Rescale Signal     event weights by factor: 1
--- DataSetFactory           : --> Rescale Background event weights by factor: 2.17066
--- DataSetFactory           : Number of training and testing events after rescaling:
--- DataSetFactory           : ------------------------------------------------------
--- DataSetFactory           : Signal     -- training events            : 25757 (sum of weights: 25757) - requested were 0 events
--- DataSetFactory           : Signal     -- testing events             : 25757 (sum of weights: 25757) - requested were 0 events
--- DataSetFactory           : Signal     -- training and testing events: 51514 (sum of weights: 51514)
--- DataSetFactory           : Background -- training events            : 11866 (sum of weights: 25757) - requested were 0 events
--- DataSetFactory           : Background -- testing events             : 11866 (sum of weights: 11866) - requested were 0 events
--- DataSetFactory           : Background -- training and testing events: 23732 (sum of weights: 37623)
--- DataSetFactory           : Create internal training tree
--- DataSetFactory           : Create internal testing tree
--- DataSetInfo              : Correlation matrix (Signal):
--- DataSetInfo              : -------------------------------------------------------------------------------------------
--- DataSetInfo              :                mult log_partPt log_partP log_ptB log_IPs partlcs  PIDNNm ghostProb log_IPPU
--- DataSetInfo              :       mult:  +1.000     +0.016    +0.018  +0.067  +0.027  +0.110  -0.135    +0.093   -0.026
--- DataSetInfo              : log_partPt:  +0.016     +1.000    +0.486  +0.040  +0.006  -0.062  +0.357    -0.069   +0.068
--- DataSetInfo              :  log_partP:  +0.018     +0.486    +1.000  +0.009  -0.068  -0.042  +0.170    +0.011   -0.143
--- DataSetInfo              :    log_ptB:  +0.067     +0.040    +0.009  +1.000  -0.002  +0.023  -0.017    +0.024   -0.002
--- DataSetInfo              :    log_IPs:  +0.027     +0.006    -0.068  -0.002  +1.000  -0.054  +0.118    -0.067   -0.030
--- DataSetInfo              :    partlcs:  +0.110     -0.062    -0.042  +0.023  -0.054  +1.000  -0.401    +0.623   -0.071
--- DataSetInfo              :     PIDNNm:  -0.135     +0.357    +0.170  -0.017  +0.118  -0.401  +1.000    -0.430   +0.094
--- DataSetInfo              :  ghostProb:  +0.093     -0.069    +0.011  +0.024  -0.067  +0.623  -0.430    +1.000   -0.085
--- DataSetInfo              :   log_IPPU:  -0.026     +0.068    -0.143  -0.002  -0.030  -0.071  +0.094    -0.085   +1.000
--- DataSetInfo              : -------------------------------------------------------------------------------------------
--- DataSetInfo              : Correlation matrix (Background):
--- DataSetInfo              : -------------------------------------------------------------------------------------------
--- DataSetInfo              :                mult log_partPt log_partP log_ptB log_IPs partlcs  PIDNNm ghostProb log_IPPU
--- DataSetInfo              :       mult:  +1.000     -0.005    +0.034  +0.073  -0.058  +0.106  -0.159    +0.120   -0.027
--- DataSetInfo              : log_partPt:  -0.005     +1.000    +0.464  +0.053  +0.056  -0.045  +0.299    -0.038   +0.046
--- DataSetInfo              :  log_partP:  +0.034     +0.464    +1.000  +0.014  -0.021  -0.067  +0.125    +0.011   -0.134
--- DataSetInfo              :    log_ptB:  +0.073     +0.053    +0.014  +1.000  -0.017  +0.018  -0.021    +0.019   +0.004
--- DataSetInfo              :    log_IPs:  -0.058     +0.056    -0.021  -0.017  +1.000  -0.120  +0.236    -0.128   -0.082
--- DataSetInfo              :    partlcs:  +0.106     -0.045    -0.067  +0.018  -0.120  +1.000  -0.428    +0.662   -0.040
--- DataSetInfo              :     PIDNNm:  -0.159     +0.299    +0.125  -0.021  +0.236  -0.428  +1.000    -0.435   +0.083
--- DataSetInfo              :  ghostProb:  +0.120     -0.038    +0.011  +0.019  -0.128  +0.662  -0.435    +1.000   -0.067
--- DataSetInfo              :   log_IPPU:  -0.027     +0.046    -0.134  +0.004  -0.082  -0.040  +0.083    -0.067   +1.000
--- DataSetInfo              : -------------------------------------------------------------------------------------------
--- DataSetFactory           :  
--- Factory                  : 
--- Factory                  : current transformation string: 'I,D,N'
--- Factory                  : Create Transformation "I" with events from all classes.
--- Id                       : Transformation, Variable selection : 
--- Id                       : Input : variable 'mult' (index=0).   <---> Output : variable 'mult' (index=0).
--- Id                       : Input : variable 'log_partPt' (index=1).   <---> Output : variable 'log_partPt' (index=1).
--- Id                       : Input : variable 'log_partP' (index=2).   <---> Output : variable 'log_partP' (index=2).
--- Id                       : Input : variable 'log_ptB' (index=3).   <---> Output : variable 'log_ptB' (index=3).
--- Id                       : Input : variable 'log_IPs' (index=4).   <---> Output : variable 'log_IPs' (index=4).
--- Id                       : Input : variable 'partlcs' (index=5).   <---> Output : variable 'partlcs' (index=5).
--- Id                       : Input : variable 'PIDNNm' (index=6).   <---> Output : variable 'PIDNNm' (index=6).
--- Id                       : Input : variable 'ghostProb' (index=7).   <---> Output : variable 'ghostProb' (index=7).
--- Id                       : Input : variable 'log_IPPU' (index=8).   <---> Output : variable 'log_IPPU' (index=8).
--- Factory                  : Create Transformation "D" with events from all classes.
--- Deco                     : Transformation, Variable selection : 
--- Deco                     : Input : variable 'mult' (index=0).   <---> Output : variable 'mult' (index=0).
--- Deco                     : Input : variable 'log_partPt' (index=1).   <---> Output : variable 'log_partPt' (index=1).
--- Deco                     : Input : variable 'log_partP' (index=2).   <---> Output : variable 'log_partP' (index=2).
--- Deco                     : Input : variable 'log_ptB' (index=3).   <---> Output : variable 'log_ptB' (index=3).
--- Deco                     : Input : variable 'log_IPs' (index=4).   <---> Output : variable 'log_IPs' (index=4).
--- Deco                     : Input : variable 'partlcs' (index=5).   <---> Output : variable 'partlcs' (index=5).
--- Deco                     : Input : variable 'PIDNNm' (index=6).   <---> Output : variable 'PIDNNm' (index=6).
--- Deco                     : Input : variable 'ghostProb' (index=7).   <---> Output : variable 'ghostProb' (index=7).
--- Deco                     : Input : variable 'log_IPPU' (index=8).   <---> Output : variable 'log_IPPU' (index=8).
--- Factory                  : Create Transformation "N" with events from all classes.
--- Norm                     : Transformation, Variable selection : 
--- Norm                     : Input : variable 'mult' (index=0).   <---> Output : variable 'mult' (index=0).
--- Norm                     : Input : variable 'log_partPt' (index=1).   <---> Output : variable 'log_partPt' (index=1).
--- Norm                     : Input : variable 'log_partP' (index=2).   <---> Output : variable 'log_partP' (index=2).
--- Norm                     : Input : variable 'log_ptB' (index=3).   <---> Output : variable 'log_ptB' (index=3).
--- Norm                     : Input : variable 'log_IPs' (index=4).   <---> Output : variable 'log_IPs' (index=4).
--- Norm                     : Input : variable 'partlcs' (index=5).   <---> Output : variable 'partlcs' (index=5).
--- Norm                     : Input : variable 'PIDNNm' (index=6).   <---> Output : variable 'PIDNNm' (index=6).
--- Norm                     : Input : variable 'ghostProb' (index=7).   <---> Output : variable 'ghostProb' (index=7).
--- Norm                     : Input : variable 'log_IPPU' (index=8).   <---> Output : variable 'log_IPPU' (index=8).
--- Id                       : Preparing the Identity transformation...
--- Deco                     : Preparing the Decorrelation transformation...
--- Norm                     : Preparing the transformation.
--- TFHandler_Factory        : ---------------------------------------------------------------------
--- TFHandler_Factory        :   Variable          Mean          RMS   [        Min          Max ]
--- TFHandler_Factory        : ---------------------------------------------------------------------
--- TFHandler_Factory        :       mult:    -0.52880     0.22636   [     -1.0000      1.0000 ]
--- TFHandler_Factory        : log_partPt:    -0.23484     0.31826   [     -1.0000      1.0000 ]
--- TFHandler_Factory        :  log_partP:   -0.074970     0.35720   [     -1.0000      1.0000 ]
--- TFHandler_Factory        :    log_ptB:     0.29888     0.18839   [     -1.0000      1.0000 ]
--- TFHandler_Factory        :    log_IPs:     0.16522     0.22385   [     -1.0000      1.0000 ]
--- TFHandler_Factory        :    partlcs:    -0.34323     0.23390   [     -1.0000      1.0000 ]
--- TFHandler_Factory        :     PIDNNm:     0.15345     0.30579   [     -1.0000      1.0000 ]
--- TFHandler_Factory        :  ghostProb:    -0.66034     0.12437   [     -1.0000      1.0000 ]
--- TFHandler_Factory        :   log_IPPU:     0.11455     0.52746   [     -1.0000      1.0000 ]
--- TFHandler_Factory        : ---------------------------------------------------------------------
--- TFHandler_Factory        : Plot event variables for Id_Deco_Norm
--- TFHandler_Factory        : Create scatter and profile plots in target-file directory: 
--- TFHandler_Factory        : /mnt/mfs/notebook/analyses/tagging_LHCb/MC/tmp4lfJ0l/result.root:/InputVariables_Id_Deco_Norm/CorrelationPlots
--- TFHandler_Factory        :  
--- TFHandler_Factory        : Ranking input variables (method unspecific)...
--- Id_Deco_NormTransforma...: Ranking result (top variable is best ranked)
--- Id_Deco_NormTransforma...: -----------------------------------
--- Id_Deco_NormTransforma...: Rank : Variable   : Separation
--- Id_Deco_NormTransforma...: -----------------------------------
--- Id_Deco_NormTransforma...:    1 : log_partPt : 1.017e-02
--- Id_Deco_NormTransforma...:    2 : PIDNNm     : 8.084e-03
--- Id_Deco_NormTransforma...:    3 : log_partP  : 7.030e-03
--- Id_Deco_NormTransforma...:    4 : log_IPs    : 5.773e-03
--- Id_Deco_NormTransforma...:    5 : mult       : 4.112e-03
--- Id_Deco_NormTransforma...:    6 : ghostProb  : 3.314e-03
--- Id_Deco_NormTransforma...:    7 : partlcs    : 2.516e-03
--- Id_Deco_NormTransforma...:    8 : log_IPPU   : 2.088e-03
--- Id_Deco_NormTransforma...:    9 : log_ptB    : 1.903e-03
--- Id_Deco_NormTransforma...: -----------------------------------
--- Factory                  :  
--- Factory                  : Train all methods for Classification ...
--- Factory                  : Train method: REP_Estimator for Classification
--- REP_Estimator            : Begin training
--- REP_Estimator            : Training Network
Error in <TDecompLU::InvertLU>: matrix is singular, 17 diag elements < tolerance of 2.2204e-16
--- <WARNING> REP_Estimator            : Line search increased error! Something is wrong.fLastAlpha=4.27732al123=2 6 18 err1=35408.3 errfinal=35411.5
--- <WARNING> REP_Estimator            : 
--- <WARNING> REP_Estimator            : negative dError=-4.64661e-05
--- REP_Estimator            : Finalizing handling of Regulator terms, trainE=1.35779 testE=1.35812
--- REP_Estimator            : Done with handling of Regulator terms
--- REP_Estimator            : End of training                                              
--- REP_Estimator            : Elapsed time for training with 37623 events: 392 sec         
--- REP_Estimator            : Create MVA output for classification on training sample
--- REP_Estimator            : Evaluation of REP_Estimator on training sample (37623 events)
--- REP_Estimator            : Elapsed time for evaluation of 37623 events: 0.139 sec       
--- REP_Estimator            : Creating weight file in xml format: weights/TMVAEstimation_REP_Estimator.weights.xml
--- REP_Estimator            : Creating standalone response class: weights/TMVAEstimation_REP_Estimator.class.C
--- REP_Estimator            : Write special histos to file: /mnt/mfs/notebook/analyses/tagging_LHCb/MC/tmp4lfJ0l/result.root:/Method_MLP/REP_Estimator
--- Factory                  : Training finished
--- Factory                  : 
--- Factory                  : Ranking input variables (method specific)...
--- REP_Estimator            : Ranking result (top variable is best ranked)
--- REP_Estimator            : -----------------------------------
--- REP_Estimator            : Rank : Variable   : Importance
--- REP_Estimator            : -----------------------------------
--- REP_Estimator            :    1 : mult       : 2.057e+04
--- REP_Estimator            :    2 : log_IPPU   : 5.486e+02
--- REP_Estimator            :    3 : log_partP  : 1.207e+02
--- REP_Estimator            :    4 : log_IPs    : 5.196e+01
--- REP_Estimator            :    5 : log_ptB    : 4.859e+01
--- REP_Estimator            :    6 : partlcs    : 2.031e+01
--- REP_Estimator            :    7 : PIDNNm     : 1.221e+01
--- REP_Estimator            :    8 : log_partPt : 5.004e+00
--- REP_Estimator            :    9 : ghostProb  : 2.288e-02
--- REP_Estimator            : -----------------------------------
--- Factory                  : 
--- Factory                  : === Destroy and recreate all methods via weight files for testing ===
--- Factory                  : 
--- MethodBase               : Reading weight file: weights/TMVAEstimation_REP_Estimator.weights.xml
--- REP_Estimator            : Read method "REP_Estimator" of type "MLP"
--- REP_Estimator            : MVA method was trained with TMVA Version: 4.2.0
--- REP_Estimator            : MVA method was trained with ROOT Version: 5.34/32
--- REP_Estimator            : Building Network
--- REP_Estimator            : Initializing weights

#error "You need a ISO C conforming compiler to use the glibc headers"
*** Interpreter error recovered ***
--- Factory                  : You are running ROOT Version: 5.34/32, Jun 23, 2015
--- Factory                  : 
--- Factory                  : _/_/_/_/_/ _|      _|  _|      _|    _|_|   
--- Factory                  :    _/      _|_|  _|_|  _|      _|  _|    _| 
--- Factory                  :   _/       _|  _|  _|  _|      _|  _|_|_|_| 
--- Factory                  :  _/        _|      _|    _|  _|    _|    _| 
--- Factory                  : _/         _|      _|      _|      _|    _| 
--- Factory                  : 
--- Factory                  : ___________TMVA Version 4.2.0, Sep 19, 2013
--- Factory                  : 
--- DataSetInfo              : Added class "Background"	 with internal class number 0
--- DataSetInfo              : Added class "Signal"	 with internal class number 1
--- Factory                  : Add Tree TrainAssignTree_Background of type Background with 32967 events
--- Factory                  : Add Tree TestAssignTree_Background of type Background with 32967 events
--- Factory                  : Add Tree TrainAssignTree_Signal of type Signal with 58339 events
--- Factory                  : Add Tree TestAssignTree_Signal of type Signal with 58339 events
--- DataSetInfo              : Class index : 0  name : Background
--- DataSetInfo              : Class index : 1  name : Signal
--- Factory                  : Booking method: REP_Estimator
--- REP_Estimator            : Building Network
--- REP_Estimator            : Initializing weights
--- DataSetFactory           : Splitmode is: "RANDOM" the mixmode is: "SAMEASSPLITMODE"
--- DataSetFactory           : Create training and testing trees -- looping over class "Background" ...
--- DataSetFactory           : Weight expression for class 'Background': "weight"
--- DataSetFactory           : Create training and testing trees -- looping over class "Signal" ...
--- DataSetFactory           : Weight expression for class 'Signal': "weight"
--- DataSetFactory           : Number of events in input trees (after possible flattening of arrays):
--- DataSetFactory           :     Background      -- number of events       : 65934  / sum of weights: 65934
--- DataSetFactory           :     Signal          -- number of events       : 116678  / sum of weights: 116678
--- DataSetFactory           :     Background tree -- total number of entries: 65934
--- DataSetFactory           :     Signal     tree -- total number of entries: 116678
--- DataSetFactory           : Preselection: (will NOT affect number of requested training and testing events)
--- DataSetFactory           :     Background requirement: "1"
--- DataSetFactory           :     Background      -- number of events passed: 65934  / sum of weights: 65934
--- DataSetFactory           :     Background      -- efficiency             : 1     
--- DataSetFactory           :     Signal     requirement: "1"
--- DataSetFactory           :     Signal          -- number of events passed: 116678  / sum of weights: 116678
--- DataSetFactory           :     Signal          -- efficiency             : 1     
--- DataSetFactory           : Weight renormalisation mode: "EqualNumEvents": renormalises all event classes ...
--- DataSetFactory           :  such that the effective (weighted) number of events in each class is the same 
--- DataSetFactory           :  (and equals the number of events (entries) given for class=0 )
--- DataSetFactory           : ... i.e. such that Sum[i=1..N_j]{w_i} = N_classA, j=classA, classB, ...
--- DataSetFactory           : ... (note that N_j is the sum of TRAINING events
--- DataSetFactory           :  ..... Testing events are not renormalised nor included in the renormalisation factor!)
--- DataSetFactory           : --> Rescale Background event weights by factor: 1
--- DataSetFactory           : --> Rescale Signal     event weights by factor: 0.565094
--- DataSetFactory           : Number of training and testing events after rescaling:
--- DataSetFactory           : ------------------------------------------------------
--- DataSetFactory           : Background -- training events            : 32967 (sum of weights: 32967) - requested were 0 events
--- DataSetFactory           : Background -- testing events             : 32967 (sum of weights: 32967) - requested were 0 events
--- DataSetFactory           : Background -- training and testing events: 65934 (sum of weights: 65934)
--- DataSetFactory           : Signal     -- training events            : 58339 (sum of weights: 32967) - requested were 0 events
--- DataSetFactory           : Signal     -- testing events             : 58339 (sum of weights: 58339) - requested were 0 events
--- DataSetFactory           : Signal     -- training and testing events: 116678 (sum of weights: 91306)
--- DataSetFactory           : Create internal training tree
--- DataSetFactory           : Create internal testing tree
--- DataSetInfo              : Correlation matrix (Background):
--- DataSetInfo              : ----------------------------------------------------------------------------------------------------------------------
--- DataSetInfo              :                 mult  nnkrec log_ptB   vflag log_ipsmean log_ptmean vcharge log_svm log_svp BDphiDir log_svtau docamax
--- DataSetInfo              :        mult:  +1.000  +0.086  +0.070  +0.217      -0.071     -0.118  +0.018  +0.202  +0.127   +0.003    -0.068  +0.220
--- DataSetInfo              :      nnkrec:  +0.086  +1.000  +0.006  +0.029      +0.037     -0.030  +0.037  +0.075  +0.038   -0.007    +0.010  +0.058
--- DataSetInfo              :     log_ptB:  +0.070  +0.006  +1.000  +0.012      +0.008     +0.064  -0.008  +0.004  +0.038   +0.006    -0.019  +0.009
--- DataSetInfo              :       vflag:  +0.217  +0.029  +0.012  +1.000      -0.079     -0.463  -0.273  +0.540  +0.258   +0.001    +0.080  +0.524
--- DataSetInfo              : log_ipsmean:  -0.071  +0.037  +0.008  -0.079      +1.000     +0.233  -0.026  -0.018  -0.088   -0.006    +0.650  -0.183
--- DataSetInfo              :  log_ptmean:  -0.118  -0.030  +0.064  -0.463      +0.233     +1.000  +0.099  -0.121  +0.171   +0.001    -0.086  -0.423
--- DataSetInfo              :     vcharge:  +0.018  +0.037  -0.008  -0.273      -0.026     +0.099  +1.000  -0.097  -0.038   -0.003    -0.062  -0.089
--- DataSetInfo              :     log_svm:  +0.202  +0.075  +0.004  +0.540      -0.018     -0.121  -0.097  +1.000  +0.347   +0.002    -0.057  +0.408
--- DataSetInfo              :     log_svp:  +0.127  +0.038  +0.038  +0.258      -0.088     +0.171  -0.038  +0.347  +1.000   +0.002    -0.124  +0.159
--- DataSetInfo              :    BDphiDir:  +0.003  -0.007  +0.006  +0.001      -0.006     +0.001  -0.003  +0.002  +0.002   +1.000    -0.009  -0.006
--- DataSetInfo              :   log_svtau:  -0.068  +0.010  -0.019  +0.080      +0.650     -0.086  -0.062  -0.057  -0.124   -0.009    +1.000  -0.046
--- DataSetInfo              :     docamax:  +0.220  +0.058  +0.009  +0.524      -0.183     -0.423  -0.089  +0.408  +0.159   -0.006    -0.046  +1.000
--- DataSetInfo              : ----------------------------------------------------------------------------------------------------------------------
--- DataSetInfo              : Correlation matrix (Signal):
--- DataSetInfo              : ----------------------------------------------------------------------------------------------------------------------
--- DataSetInfo              :                 mult  nnkrec log_ptB   vflag log_ipsmean log_ptmean vcharge log_svm log_svp BDphiDir log_svtau docamax
--- DataSetInfo              :        mult:  +1.000  +0.087  +0.068  +0.219      -0.046     -0.118  -0.033  +0.167  +0.107   +0.003    -0.051  +0.208
--- DataSetInfo              :      nnkrec:  +0.087  +1.000  +0.003  +0.024      +0.016     -0.037  +0.020  +0.051  +0.022   -0.004    +0.014  +0.058
--- DataSetInfo              :     log_ptB:  +0.068  +0.003  +1.000  +0.013      +0.017     +0.068  -0.004  +0.006  +0.033   -0.003    -0.010  +0.005
--- DataSetInfo              :       vflag:  +0.219  +0.024  +0.013  +1.000      -0.065     -0.438  -0.334  +0.529  +0.263   -0.001    +0.097  +0.514
--- DataSetInfo              : log_ipsmean:  -0.046  +0.016  +0.017  -0.065      +1.000     +0.212  -0.035  +0.010  -0.073   -0.002    +0.657  -0.174
--- DataSetInfo              :  log_ptmean:  -0.118  -0.037  +0.068  -0.438      +0.212     +1.000  +0.169  -0.068  +0.210   +0.004    -0.135  -0.399
--- DataSetInfo              :     vcharge:  -0.033  +0.020  -0.004  -0.334      -0.035     +0.169  +1.000  -0.106  -0.057   -0.000    -0.096  -0.140
--- DataSetInfo              :     log_svm:  +0.167  +0.051  +0.006  +0.529      +0.010     -0.068  -0.106  +1.000  +0.342   +0.002    -0.030  +0.378
--- DataSetInfo              :     log_svp:  +0.107  +0.022  +0.033  +0.263      -0.073     +0.210  -0.057  +0.342  +1.000   +0.001    -0.106  +0.146
--- DataSetInfo              :    BDphiDir:  +0.003  -0.004  -0.003  -0.001      -0.002     +0.004  -0.000  +0.002  +0.001   +1.000    -0.005  -0.005
--- DataSetInfo              :   log_svtau:  -0.051  +0.014  -0.010  +0.097      +0.657     -0.135  -0.096  -0.030  -0.106   -0.005    +1.000  -0.026
--- DataSetInfo              :     docamax:  +0.208  +0.058  +0.005  +0.514      -0.174     -0.399  -0.140  +0.378  +0.146   -0.005    -0.026  +1.000
--- DataSetInfo              : ----------------------------------------------------------------------------------------------------------------------
--- DataSetFactory           :  
--- Factory                  : 
--- Factory                  : current transformation string: 'I,D,N'
--- Factory                  : Create Transformation "I" with events from all classes.
--- Id                       : Transformation, Variable selection : 
--- Id                       : Input : variable 'mult' (index=0).   <---> Output : variable 'mult' (index=0).
--- Id                       : Input : variable 'nnkrec' (index=1).   <---> Output : variable 'nnkrec' (index=1).
--- Id                       : Input : variable 'log_ptB' (index=2).   <---> Output : variable 'log_ptB' (index=2).
--- Id                       : Input : variable 'vflag' (index=3).   <---> Output : variable 'vflag' (index=3).
--- Id                       : Input : variable 'log_ipsmean' (index=4).   <---> Output : variable 'log_ipsmean' (index=4).
--- Id                       : Input : variable 'log_ptmean' (index=5).   <---> Output : variable 'log_ptmean' (index=5).
--- Id                       : Input : variable 'vcharge' (index=6).   <---> Output : variable 'vcharge' (index=6).
--- Id                       : Input : variable 'log_svm' (index=7).   <---> Output : variable 'log_svm' (index=7).
--- Id                       : Input : variable 'log_svp' (index=8).   <---> Output : variable 'log_svp' (index=8).
--- Id                       : Input : variable 'BDphiDir' (index=9).   <---> Output : variable 'BDphiDir' (index=9).
--- Id                       : Input : variable 'log_svtau' (index=10).   <---> Output : variable 'log_svtau' (index=10).
--- Id                       : Input : variable 'docamax' (index=11).   <---> Output : variable 'docamax' (index=11).
--- Factory                  : Create Transformation "D" with events from all classes.
--- Deco                     : Transformation, Variable selection : 
--- Deco                     : Input : variable 'mult' (index=0).   <---> Output : variable 'mult' (index=0).
--- Deco                     : Input : variable 'nnkrec' (index=1).   <---> Output : variable 'nnkrec' (index=1).
--- Deco                     : Input : variable 'log_ptB' (index=2).   <---> Output : variable 'log_ptB' (index=2).
--- Deco                     : Input : variable 'vflag' (index=3).   <---> Output : variable 'vflag' (index=3).
--- Deco                     : Input : variable 'log_ipsmean' (index=4).   <---> Output : variable 'log_ipsmean' (index=4).
--- Deco                     : Input : variable 'log_ptmean' (index=5).   <---> Output : variable 'log_ptmean' (index=5).
--- Deco                     : Input : variable 'vcharge' (index=6).   <---> Output : variable 'vcharge' (index=6).
--- Deco                     : Input : variable 'log_svm' (index=7).   <---> Output : variable 'log_svm' (index=7).
--- Deco                     : Input : variable 'log_svp' (index=8).   <---> Output : variable 'log_svp' (index=8).
--- Deco                     : Input : variable 'BDphiDir' (index=9).   <---> Output : variable 'BDphiDir' (index=9).
--- Deco                     : Input : variable 'log_svtau' (index=10).   <---> Output : variable 'log_svtau' (index=10).
--- Deco                     : Input : variable 'docamax' (index=11).   <---> Output : variable 'docamax' (index=11).
--- Factory                  : Create Transformation "N" with events from all classes.
--- Norm                     : Transformation, Variable selection : 
--- Norm                     : Input : variable 'mult' (index=0).   <---> Output : variable 'mult' (index=0).
--- Norm                     : Input : variable 'nnkrec' (index=1).   <---> Output : variable 'nnkrec' (index=1).
--- Norm                     : Input : variable 'log_ptB' (index=2).   <---> Output : variable 'log_ptB' (index=2).
--- Norm                     : Input : variable 'vflag' (index=3).   <---> Output : variable 'vflag' (index=3).
--- Norm                     : Input : variable 'log_ipsmean' (index=4).   <---> Output : variable 'log_ipsmean' (index=4).
--- Norm                     : Input : variable 'log_ptmean' (index=5).   <---> Output : variable 'log_ptmean' (index=5).
--- Norm                     : Input : variable 'vcharge' (index=6).   <---> Output : variable 'vcharge' (index=6).
--- Norm                     : Input : variable 'log_svm' (index=7).   <---> Output : variable 'log_svm' (index=7).
--- Norm                     : Input : variable 'log_svp' (index=8).   <---> Output : variable 'log_svp' (index=8).
--- Norm                     : Input : variable 'BDphiDir' (index=9).   <---> Output : variable 'BDphiDir' (index=9).
--- Norm                     : Input : variable 'log_svtau' (index=10).   <---> Output : variable 'log_svtau' (index=10).
--- Norm                     : Input : variable 'docamax' (index=11).   <---> Output : variable 'docamax' (index=11).
--- Id                       : Preparing the Identity transformation...
--- Deco                     : Preparing the Decorrelation transformation...
--- Norm                     : Preparing the transformation.
--- TFHandler_Factory        : --------------------------------------------------------------------------
--- TFHandler_Factory        :    Variable           Mean           RMS   [        Min           Max ]
--- TFHandler_Factory        : --------------------------------------------------------------------------
--- TFHandler_Factory        :        mult:     -0.62244      0.18718   [      -1.0000       1.0000 ]
--- TFHandler_Factory        :      nnkrec:     -0.58050      0.26330   [      -1.0000       1.0000 ]
--- TFHandler_Factory        :     log_ptB:      0.34715      0.20779   [      -1.0000       1.0000 ]
--- TFHandler_Factory        :       vflag:     -0.55786      0.17873   [      -1.0000       1.0000 ]
--- TFHandler_Factory        : log_ipsmean:    -0.067655      0.23288   [      -1.0000       1.0000 ]
--- TFHandler_Factory        :  log_ptmean:     -0.29881      0.24866   [      -1.0000       1.0000 ]
--- TFHandler_Factory        :     vcharge:     -0.21271      0.40218   [      -1.0000       1.0000 ]
--- TFHandler_Factory        :     log_svm:     -0.14657      0.25925   [      -1.0000       1.0000 ]
--- TFHandler_Factory        :     log_svp:     -0.12375      0.31620   [      -1.0000       1.0000 ]
--- TFHandler_Factory        :    BDphiDir:    0.0017789      0.57725   [      -1.0000       1.0000 ]
--- TFHandler_Factory        :   log_svtau:     0.075845      0.16638   [      -1.0000       1.0000 ]
--- TFHandler_Factory        :     docamax:     -0.36165      0.20407   [      -1.0000       1.0000 ]
--- TFHandler_Factory        : --------------------------------------------------------------------------
--- TFHandler_Factory        : Plot event variables for Id_Deco_Norm
--- TFHandler_Factory        : Create scatter and profile plots in target-file directory: 
--- TFHandler_Factory        : /mnt/mfs/notebook/analyses/tagging_LHCb/MC/tmps6WtPC/result.root:/InputVariables_Id_Deco_Norm/CorrelationPlots
--- TFHandler_Factory        :  
--- TFHandler_Factory        : Ranking input variables (method unspecific)...
--- Id_Deco_NormTransforma...: Ranking result (top variable is best ranked)
--- Id_Deco_NormTransforma...: ------------------------------------
--- Id_Deco_NormTransforma...: Rank : Variable    : Separation
--- Id_Deco_NormTransforma...: ------------------------------------
--- Id_Deco_NormTransforma...:    1 : vcharge     : 1.010e-02
--- Id_Deco_NormTransforma...:    2 : log_ptmean  : 4.391e-03
--- Id_Deco_NormTransforma...:    3 : log_svtau   : 2.803e-03
--- Id_Deco_NormTransforma...:    4 : mult        : 2.732e-03
--- Id_Deco_NormTransforma...:    5 : log_svm     : 2.707e-03
--- Id_Deco_NormTransforma...:    6 : nnkrec      : 1.646e-03
--- Id_Deco_NormTransforma...:    7 : docamax     : 1.390e-03
--- Id_Deco_NormTransforma...:    8 : vflag       : 1.360e-03
--- Id_Deco_NormTransforma...:    9 : log_ipsmean : 1.149e-03
--- Id_Deco_NormTransforma...:   10 : log_svp     : 8.720e-04
--- Id_Deco_NormTransforma...:   11 : log_ptB     : 5.962e-04
--- Id_Deco_NormTransforma...:   12 : BDphiDir    : 4.004e-04
--- Id_Deco_NormTransforma...: ------------------------------------
--- Factory                  :  
--- Factory                  : Train all methods for Classification ...
--- Factory                  : Train method: REP_Estimator for Classification
--- REP_Estimator            : Begin training
--- REP_Estimator            : Training Network
Error in <TDecompLU::InvertLU>: matrix is singular, 10 diag elements < tolerance of 2.2204e-16
Error in <TDecompLU::InvertLU>: matrix is singular, 4 diag elements < tolerance of 2.2204e-16
Error in <TDecompLU::InvertLU>: matrix is singular, 2 diag elements < tolerance of 2.2204e-16
Error in <TDecompLU::InvertLU>: matrix is singular, 5 diag elements < tolerance of 2.2204e-16
--- REP_Estimator            : Finalizing handling of Regulator terms, trainE=1.37124 testE=1.37232
--- REP_Estimator            : Done with handling of Regulator terms
--- REP_Estimator            : End of training                                              
--- REP_Estimator            : Elapsed time for training with 91306 events: 905 sec         
--- REP_Estimator            : Create MVA output for classification on training sample
--- REP_Estimator            : Evaluation of REP_Estimator on training sample (91306 events)
--- REP_Estimator            : Elapsed time for evaluation of 91306 events: 0.288 sec       
--- REP_Estimator            : Creating weight file in xml format: weights/TMVAEstimation_REP_Estimator.weights.xml
--- REP_Estimator            : Creating standalone response class: weights/TMVAEstimation_REP_Estimator.class.C
--- REP_Estimator            : Write special histos to file: /mnt/mfs/notebook/analyses/tagging_LHCb/MC/tmps6WtPC/result.root:/Method_MLP/REP_Estimator
--- Factory                  : Training finished
--- Factory                  : 
--- Factory                  : Ranking input variables (method specific)...
--- REP_Estimator            : Ranking result (top variable is best ranked)
--- REP_Estimator            : ------------------------------------
--- REP_Estimator            : Rank : Variable    : Importance
--- REP_Estimator            : ------------------------------------
--- REP_Estimator            :    1 : mult        : 5.931e+04
--- REP_Estimator            :    2 : log_svp     : 3.959e+02
--- REP_Estimator            :    3 : vflag       : 2.636e+02
--- REP_Estimator            :    4 : nnkrec      : 9.831e+01
--- REP_Estimator            :    5 : log_ipsmean : 8.159e+01
--- REP_Estimator            :    6 : log_ptB     : 5.000e+01
--- REP_Estimator            :    7 : BDphiDir    : 1.986e+01
--- REP_Estimator            :    8 : vcharge     : 1.504e+01
--- REP_Estimator            :    9 : log_svm     : 1.336e+01
--- REP_Estimator            :   10 : log_svtau   : 1.232e+01
--- REP_Estimator            :   11 : log_ptmean  : 5.607e+00
--- REP_Estimator            :   12 : docamax     : 1.024e-02
--- REP_Estimator            : ------------------------------------
--- Factory                  : 
--- Factory                  : === Destroy and recreate all methods via weight files for testing ===
--- Factory                  : 
--- MethodBase               : Reading weight file: weights/TMVAEstimation_REP_Estimator.weights.xml
--- REP_Estimator            : Read method "REP_Estimator" of type "MLP"
--- REP_Estimator            : MVA method was trained with TMVA Version: 4.2.0
--- REP_Estimator            : MVA method was trained with ROOT Version: 5.34/32
--- REP_Estimator            : Building Network
--- REP_Estimator            : Initializing weights

#error "You need a ISO C conforming compiler to use the glibc headers"
*** Interpreter error recovered ***
--- Factory                  : You are running ROOT Version: 5.34/32, Jun 23, 2015
--- Factory                  : 
--- Factory                  : _/_/_/_/_/ _|      _|  _|      _|    _|_|   
--- Factory                  :    _/      _|_|  _|_|  _|      _|  _|    _| 
--- Factory                  :   _/       _|  _|  _|  _|      _|  _|_|_|_| 
--- Factory                  :  _/        _|      _|    _|  _|    _|    _| 
--- Factory                  : _/         _|      _|      _|      _|    _| 
--- Factory                  : 
--- Factory                  : ___________TMVA Version 4.2.0, Sep 19, 2013
--- Factory                  : 
--- DataSetInfo              : Added class "Signal"	 with internal class number 0
--- DataSetInfo              : Added class "Background"	 with internal class number 1
--- Factory                  : Add Tree TrainAssignTree_Signal of type Signal with 58343 events
--- Factory                  : Add Tree TestAssignTree_Signal of type Signal with 58343 events
--- Factory                  : Add Tree TrainAssignTree_Background of type Background with 32964 events
--- Factory                  : Add Tree TestAssignTree_Background of type Background with 32964 events
--- DataSetInfo              : Class index : 0  name : Signal
--- DataSetInfo              : Class index : 1  name : Background
--- Factory                  : Booking method: REP_Estimator
--- REP_Estimator            : Building Network
--- REP_Estimator            : Initializing weights
--- DataSetFactory           : Splitmode is: "RANDOM" the mixmode is: "SAMEASSPLITMODE"
--- DataSetFactory           : Create training and testing trees -- looping over class "Signal" ...
--- DataSetFactory           : Weight expression for class 'Signal': "weight"
--- DataSetFactory           : Create training and testing trees -- looping over class "Background" ...
--- DataSetFactory           : Weight expression for class 'Background': "weight"
--- DataSetFactory           : Number of events in input trees (after possible flattening of arrays):
--- DataSetFactory           :     Signal          -- number of events       : 116686  / sum of weights: 116686
--- DataSetFactory           :     Background      -- number of events       : 65928  / sum of weights: 65928
--- DataSetFactory           :     Signal     tree -- total number of entries: 116686
--- DataSetFactory           :     Background tree -- total number of entries: 65928
--- DataSetFactory           : Preselection: (will NOT affect number of requested training and testing events)
--- DataSetFactory           :     Signal     requirement: "1"
--- DataSetFactory           :     Signal          -- number of events passed: 116686  / sum of weights: 116686
--- DataSetFactory           :     Signal          -- efficiency             : 1     
--- DataSetFactory           :     Background requirement: "1"
--- DataSetFactory           :     Background      -- number of events passed: 65928  / sum of weights: 65928
--- DataSetFactory           :     Background      -- efficiency             : 1     
--- DataSetFactory           : Weight renormalisation mode: "EqualNumEvents": renormalises all event classes ...
--- DataSetFactory           :  such that the effective (weighted) number of events in each class is the same 
--- DataSetFactory           :  (and equals the number of events (entries) given for class=0 )
--- DataSetFactory           : ... i.e. such that Sum[i=1..N_j]{w_i} = N_classA, j=classA, classB, ...
--- DataSetFactory           : ... (note that N_j is the sum of TRAINING events
--- DataSetFactory           :  ..... Testing events are not renormalised nor included in the renormalisation factor!)
--- DataSetFactory           : --> Rescale Signal     event weights by factor: 1
--- DataSetFactory           : --> Rescale Background event weights by factor: 1.7699
--- DataSetFactory           : Number of training and testing events after rescaling:
--- DataSetFactory           : ------------------------------------------------------
--- DataSetFactory           : Signal     -- training events            : 58343 (sum of weights: 58343) - requested were 0 events
--- DataSetFactory           : Signal     -- testing events             : 58343 (sum of weights: 58343) - requested were 0 events
--- DataSetFactory           : Signal     -- training and testing events: 116686 (sum of weights: 116686)
--- DataSetFactory           : Background -- training events            : 32964 (sum of weights: 58343) - requested were 0 events
--- DataSetFactory           : Background -- testing events             : 32964 (sum of weights: 32964) - requested were 0 events
--- DataSetFactory           : Background -- training and testing events: 65928 (sum of weights: 91307)
--- DataSetFactory           : Create internal training tree
--- DataSetFactory           : Create internal testing tree
--- DataSetInfo              : Correlation matrix (Signal):
--- DataSetInfo              : ----------------------------------------------------------------------------------------------------------------------
--- DataSetInfo              :                 mult  nnkrec log_ptB   vflag log_ipsmean log_ptmean vcharge log_svm log_svp BDphiDir log_svtau docamax
--- DataSetInfo              :        mult:  +1.000  +0.088  +0.063  +0.219      -0.046     -0.121  -0.025  +0.164  +0.109   -0.001    -0.044  +0.209
--- DataSetInfo              :      nnkrec:  +0.088  +1.000  +0.001  +0.027      +0.020     -0.034  +0.013  +0.054  +0.034   +0.003    +0.016  +0.065
--- DataSetInfo              :     log_ptB:  +0.063  +0.001  +1.000  +0.009      +0.007     +0.066  -0.010  +0.000  +0.045   +0.005    -0.013  +0.002
--- DataSetInfo              :       vflag:  +0.219  +0.027  +0.009  +1.000      -0.066     -0.435  -0.328  +0.527  +0.261   -0.001    +0.097  +0.516
--- DataSetInfo              : log_ipsmean:  -0.046  +0.020  +0.007  -0.066      +1.000     +0.219  -0.027  +0.022  -0.060   +0.003    +0.652  -0.176
--- DataSetInfo              :  log_ptmean:  -0.121  -0.034  +0.066  -0.435      +0.219     +1.000  +0.158  -0.063  +0.216   -0.002    -0.134  -0.402
--- DataSetInfo              :     vcharge:  -0.025  +0.013  -0.010  -0.328      -0.027     +0.158  +1.000  -0.109  -0.066   -0.002    -0.087  -0.140
--- DataSetInfo              :     log_svm:  +0.164  +0.054  +0.000  +0.527      +0.022     -0.063  -0.109  +1.000  +0.346   -0.007    -0.021  +0.374
--- DataSetInfo              :     log_svp:  +0.109  +0.034  +0.045  +0.261      -0.060     +0.216  -0.066  +0.346  +1.000   -0.002    -0.110  +0.142
--- DataSetInfo              :    BDphiDir:  -0.001  +0.003  +0.005  -0.001      +0.003     -0.002  -0.002  -0.007  -0.002   +1.000    +0.009  +0.004
--- DataSetInfo              :   log_svtau:  -0.044  +0.016  -0.013  +0.097      +0.652     -0.134  -0.087  -0.021  -0.110   +0.009    +1.000  -0.026
--- DataSetInfo              :     docamax:  +0.209  +0.065  +0.002  +0.516      -0.176     -0.402  -0.140  +0.374  +0.142   +0.004    -0.026  +1.000
--- DataSetInfo              : ----------------------------------------------------------------------------------------------------------------------
--- DataSetInfo              : Correlation matrix (Background):
--- DataSetInfo              : ----------------------------------------------------------------------------------------------------------------------
--- DataSetInfo              :                 mult  nnkrec log_ptB   vflag log_ipsmean log_ptmean vcharge log_svm log_svp BDphiDir log_svtau docamax
--- DataSetInfo              :        mult:  +1.000  +0.086  +0.063  +0.204      -0.085     -0.124  +0.012  +0.192  +0.115   +0.007    -0.077  +0.209
--- DataSetInfo              :      nnkrec:  +0.086  +1.000  +0.005  +0.023      +0.040     -0.028  +0.042  +0.079  +0.052   +0.001    +0.016  +0.070
--- DataSetInfo              :     log_ptB:  +0.063  +0.005  +1.000  +0.007      +0.006     +0.069  -0.002  -0.004  +0.036   +0.002    -0.014  -0.006
--- DataSetInfo              :       vflag:  +0.204  +0.023  +0.007  +1.000      -0.087     -0.465  -0.283  +0.541  +0.258   -0.005    +0.075  +0.515
--- DataSetInfo              : log_ipsmean:  -0.085  +0.040  +0.006  -0.087      +1.000     +0.233  -0.031  -0.018  -0.081   -0.004    +0.643  -0.184
--- DataSetInfo              :  log_ptmean:  -0.124  -0.028  +0.069  -0.465      +0.233     +1.000  +0.107  -0.120  +0.175   -0.005    -0.088  -0.413
--- DataSetInfo              :     vcharge:  +0.012  +0.042  -0.002  -0.283      -0.031     +0.107  +1.000  -0.095  -0.041   +0.002    -0.066  -0.092
--- DataSetInfo              :     log_svm:  +0.192  +0.079  -0.004  +0.541      -0.018     -0.120  -0.095  +1.000  +0.344   -0.002    -0.056  +0.410
--- DataSetInfo              :     log_svp:  +0.115  +0.052  +0.036  +0.258      -0.081     +0.175  -0.041  +0.344  +1.000   -0.012    -0.124  +0.159
--- DataSetInfo              :    BDphiDir:  +0.007  +0.001  +0.002  -0.005      -0.004     -0.005  +0.002  -0.002  -0.012   +1.000    -0.006  -0.001
--- DataSetInfo              :   log_svtau:  -0.077  +0.016  -0.014  +0.075      +0.643     -0.088  -0.066  -0.056  -0.124   -0.006    +1.000  -0.055
--- DataSetInfo              :     docamax:  +0.209  +0.070  -0.006  +0.515      -0.184     -0.413  -0.092  +0.410  +0.159   -0.001    -0.055  +1.000
--- DataSetInfo              : ----------------------------------------------------------------------------------------------------------------------
--- DataSetFactory           :  
--- Factory                  : 
--- Factory                  : current transformation string: 'I,D,N'
--- Factory                  : Create Transformation "I" with events from all classes.
--- Id                       : Transformation, Variable selection : 
--- Id                       : Input : variable 'mult' (index=0).   <---> Output : variable 'mult' (index=0).
--- Id                       : Input : variable 'nnkrec' (index=1).   <---> Output : variable 'nnkrec' (index=1).
--- Id                       : Input : variable 'log_ptB' (index=2).   <---> Output : variable 'log_ptB' (index=2).
--- Id                       : Input : variable 'vflag' (index=3).   <---> Output : variable 'vflag' (index=3).
--- Id                       : Input : variable 'log_ipsmean' (index=4).   <---> Output : variable 'log_ipsmean' (index=4).
--- Id                       : Input : variable 'log_ptmean' (index=5).   <---> Output : variable 'log_ptmean' (index=5).
--- Id                       : Input : variable 'vcharge' (index=6).   <---> Output : variable 'vcharge' (index=6).
--- Id                       : Input : variable 'log_svm' (index=7).   <---> Output : variable 'log_svm' (index=7).
--- Id                       : Input : variable 'log_svp' (index=8).   <---> Output : variable 'log_svp' (index=8).
--- Id                       : Input : variable 'BDphiDir' (index=9).   <---> Output : variable 'BDphiDir' (index=9).
--- Id                       : Input : variable 'log_svtau' (index=10).   <---> Output : variable 'log_svtau' (index=10).
--- Id                       : Input : variable 'docamax' (index=11).   <---> Output : variable 'docamax' (index=11).
--- Factory                  : Create Transformation "D" with events from all classes.
--- Deco                     : Transformation, Variable selection : 
--- Deco                     : Input : variable 'mult' (index=0).   <---> Output : variable 'mult' (index=0).
--- Deco                     : Input : variable 'nnkrec' (index=1).   <---> Output : variable 'nnkrec' (index=1).
--- Deco                     : Input : variable 'log_ptB' (index=2).   <---> Output : variable 'log_ptB' (index=2).
--- Deco                     : Input : variable 'vflag' (index=3).   <---> Output : variable 'vflag' (index=3).
--- Deco                     : Input : variable 'log_ipsmean' (index=4).   <---> Output : variable 'log_ipsmean' (index=4).
--- Deco                     : Input : variable 'log_ptmean' (index=5).   <---> Output : variable 'log_ptmean' (index=5).
--- Deco                     : Input : variable 'vcharge' (index=6).   <---> Output : variable 'vcharge' (index=6).
--- Deco                     : Input : variable 'log_svm' (index=7).   <---> Output : variable 'log_svm' (index=7).
--- Deco                     : Input : variable 'log_svp' (index=8).   <---> Output : variable 'log_svp' (index=8).
--- Deco                     : Input : variable 'BDphiDir' (index=9).   <---> Output : variable 'BDphiDir' (index=9).
--- Deco                     : Input : variable 'log_svtau' (index=10).   <---> Output : variable 'log_svtau' (index=10).
--- Deco                     : Input : variable 'docamax' (index=11).   <---> Output : variable 'docamax' (index=11).
--- Factory                  : Create Transformation "N" with events from all classes.
--- Norm                     : Transformation, Variable selection : 
--- Norm                     : Input : variable 'mult' (index=0).   <---> Output : variable 'mult' (index=0).
--- Norm                     : Input : variable 'nnkrec' (index=1).   <---> Output : variable 'nnkrec' (index=1).
--- Norm                     : Input : variable 'log_ptB' (index=2).   <---> Output : variable 'log_ptB' (index=2).
--- Norm                     : Input : variable 'vflag' (index=3).   <---> Output : variable 'vflag' (index=3).
--- Norm                     : Input : variable 'log_ipsmean' (index=4).   <---> Output : variable 'log_ipsmean' (index=4).
--- Norm                     : Input : variable 'log_ptmean' (index=5).   <---> Output : variable 'log_ptmean' (index=5).
--- Norm                     : Input : variable 'vcharge' (index=6).   <---> Output : variable 'vcharge' (index=6).
--- Norm                     : Input : variable 'log_svm' (index=7).   <---> Output : variable 'log_svm' (index=7).
--- Norm                     : Input : variable 'log_svp' (index=8).   <---> Output : variable 'log_svp' (index=8).
--- Norm                     : Input : variable 'BDphiDir' (index=9).   <---> Output : variable 'BDphiDir' (index=9).
--- Norm                     : Input : variable 'log_svtau' (index=10).   <---> Output : variable 'log_svtau' (index=10).
--- Norm                     : Input : variable 'docamax' (index=11).   <---> Output : variable 'docamax' (index=11).
--- Id                       : Preparing the Identity transformation...
--- Deco                     : Preparing the Decorrelation transformation...
--- Norm                     : Preparing the transformation.
--- TFHandler_Factory        : --------------------------------------------------------------------------
--- TFHandler_Factory        :    Variable           Mean           RMS   [        Min           Max ]
--- TFHandler_Factory        : --------------------------------------------------------------------------
--- TFHandler_Factory        :        mult:     -0.51266      0.23312   [      -1.0000       1.0000 ]
--- TFHandler_Factory        :      nnkrec:     -0.58896      0.25650   [      -1.0000       1.0000 ]
--- TFHandler_Factory        :     log_ptB:      0.25462      0.19096   [      -1.0000       1.0000 ]
--- TFHandler_Factory        :       vflag:     -0.50694      0.20497   [      -1.0000       1.0000 ]
--- TFHandler_Factory        : log_ipsmean:    -0.058246      0.20774   [      -1.0000       1.0000 ]
--- TFHandler_Factory        :  log_ptmean:     -0.33265      0.25805   [      -1.0000       1.0000 ]
--- TFHandler_Factory        :     vcharge:     -0.20425      0.41313   [      -1.0000       1.0000 ]
--- TFHandler_Factory        :     log_svm:     -0.16125      0.25009   [      -1.0000       1.0000 ]
--- TFHandler_Factory        :     log_svp:     -0.10075      0.32644   [      -1.0000       1.0000 ]
--- TFHandler_Factory        :    BDphiDir:   0.00054056      0.57774   [      -1.0000       1.0000 ]
--- TFHandler_Factory        :   log_svtau:     0.016460      0.16557   [      -1.0000       1.0000 ]
--- TFHandler_Factory        :     docamax:     -0.34878      0.20304   [      -1.0000       1.0000 ]
--- TFHandler_Factory        : --------------------------------------------------------------------------
--- TFHandler_Factory        : Plot event variables for Id_Deco_Norm
--- TFHandler_Factory        : Create scatter and profile plots in target-file directory: 
--- TFHandler_Factory        : /mnt/mfs/notebook/analyses/tagging_LHCb/MC/tmpVZK0zR/result.root:/InputVariables_Id_Deco_Norm/CorrelationPlots
--- TFHandler_Factory        :  
--- TFHandler_Factory        : Ranking input variables (method unspecific)...
--- Id_Deco_NormTransforma...: Ranking result (top variable is best ranked)
--- Id_Deco_NormTransforma...: ------------------------------------
--- Id_Deco_NormTransforma...: Rank : Variable    : Separation
--- Id_Deco_NormTransforma...: ------------------------------------
--- Id_Deco_NormTransforma...:    1 : vcharge     : 1.038e-02
--- Id_Deco_NormTransforma...:    2 : log_ptmean  : 4.576e-03
--- Id_Deco_NormTransforma...:    3 : log_svm     : 3.006e-03
--- Id_Deco_NormTransforma...:    4 : log_svtau   : 2.644e-03
--- Id_Deco_NormTransforma...:    5 : mult        : 2.414e-03
--- Id_Deco_NormTransforma...:    6 : nnkrec      : 1.579e-03
--- Id_Deco_NormTransforma...:    7 : vflag       : 1.505e-03
--- Id_Deco_NormTransforma...:    8 : log_ipsmean : 1.464e-03
--- Id_Deco_NormTransforma...:    9 : docamax     : 1.335e-03
--- Id_Deco_NormTransforma...:   10 : log_svp     : 7.404e-04
--- Id_Deco_NormTransforma...:   11 : log_ptB     : 5.202e-04
--- Id_Deco_NormTransforma...:   12 : BDphiDir    : 3.589e-04
--- Id_Deco_NormTransforma...: ------------------------------------
--- Factory                  :  
--- Factory                  : Train all methods for Classification ...
--- Factory                  : Train method: REP_Estimator for Classification
--- REP_Estimator            : Begin training
--- REP_Estimator            : Training Network
Error in <TDecompLU::InvertLU>: matrix is singular, 8 diag elements < tolerance of 2.2204e-16
Error in <TDecompLU::InvertLU>: matrix is singular, 1 diag elements < tolerance of 2.2204e-16
Error in <TDecompLU::InvertLU>: matrix is singular, 1 diag elements < tolerance of 2.2204e-16
Error in <TDecompLU::InvertLU>: matrix is singular, 1 diag elements < tolerance of 2.2204e-16
Error in <TDecompLU::InvertLU>: matrix is singular, 4 diag elements < tolerance of 2.2204e-16
Error in <TDecompLU::InvertLU>: matrix is singular, 1 diag elements < tolerance of 2.2204e-16
Error in <TDecompLU::DecomposeLUCrout>: matrix is singular
Error in <TDecompLU::InvertLU>: matrix is singular, 2 diag elements < tolerance of 2.2204e-16
--- REP_Estimator            : Finalizing handling of Regulator terms, trainE=1.36942 testE=1.37025
--- REP_Estimator            : Done with handling of Regulator terms
--- REP_Estimator            : End of training                                              
--- REP_Estimator            : Elapsed time for training with 91307 events: 939 sec         
--- REP_Estimator            : Create MVA output for classification on training sample
--- REP_Estimator            : Evaluation of REP_Estimator on training sample (91307 events)
--- REP_Estimator            : Elapsed time for evaluation of 91307 events: 0.298 sec       
--- REP_Estimator            : Creating weight file in xml format: weights/TMVAEstimation_REP_Estimator.weights.xml
--- REP_Estimator            : Creating standalone response class: weights/TMVAEstimation_REP_Estimator.class.C
--- REP_Estimator            : Write special histos to file: /mnt/mfs/notebook/analyses/tagging_LHCb/MC/tmpVZK0zR/result.root:/Method_MLP/REP_Estimator
--- Factory                  : Training finished
--- Factory                  : 
--- Factory                  : Ranking input variables (method specific)...
--- REP_Estimator            : Ranking result (top variable is best ranked)
--- REP_Estimator            : ------------------------------------
--- REP_Estimator            : Rank : Variable    : Importance
--- REP_Estimator            : ------------------------------------
--- REP_Estimator            :    1 : mult        : 8.448e+04
--- REP_Estimator            :    2 : log_svp     : 5.046e+02
--- REP_Estimator            :    3 : vflag       : 3.191e+02
--- REP_Estimator            :    4 : nnkrec      : 1.119e+02
--- REP_Estimator            :    5 : log_ipsmean : 1.025e+02
--- REP_Estimator            :    6 : log_ptB     : 5.354e+01
--- REP_Estimator            :    7 : BDphiDir    : 2.973e+01
--- REP_Estimator            :    8 : vcharge     : 1.388e+01
--- REP_Estimator            :    9 : log_svm     : 1.109e+01
--- REP_Estimator            :   10 : log_svtau   : 1.005e+01
--- REP_Estimator            :   11 : log_ptmean  : 5.149e+00
--- REP_Estimator            :   12 : docamax     : 1.086e-02
--- REP_Estimator            : ------------------------------------
--- Factory                  : 
--- Factory                  : === Destroy and recreate all methods via weight files for testing ===
--- Factory                  : 
--- MethodBase               : Reading weight file: weights/TMVAEstimation_REP_Estimator.weights.xml
--- REP_Estimator            : Read method "REP_Estimator" of type "MLP"
--- REP_Estimator            : MVA method was trained with TMVA Version: 4.2.0
--- REP_Estimator            : MVA method was trained with ROOT Version: 5.34/32
--- REP_Estimator            : Building Network
--- REP_Estimator            : Initializing weights

#error "You need a ISO C conforming compiler to use the glibc headers"
*** Interpreter error recovered ***
--- Factory                  : You are running ROOT Version: 5.34/32, Jun 23, 2015
--- Factory                  : 
--- Factory                  : _/_/_/_/_/ _|      _|  _|      _|    _|_|   
--- Factory                  :    _/      _|_|  _|_|  _|      _|  _|    _| 
--- Factory                  :   _/       _|  _|  _|  _|      _|  _|_|_|_| 
--- Factory                  :  _/        _|      _|    _|  _|    _|    _| 
--- Factory                  : _/         _|      _|      _|      _|    _| 
--- Factory                  : 
--- Factory                  : ___________TMVA Version 4.2.0, Sep 19, 2013
--- Factory                  : 
--- DataSetInfo              : Added class "Background"	 with internal class number 0
--- DataSetInfo              : Added class "Signal"	 with internal class number 1
--- Factory                  : Add Tree TrainAssignTree_Background of type Background with 45436 events
--- Factory                  : Add Tree TestAssignTree_Background of type Background with 45436 events
--- Factory                  : Add Tree TrainAssignTree_Signal of type Signal with 86775 events
--- Factory                  : Add Tree TestAssignTree_Signal of type Signal with 86775 events
--- DataSetInfo              : Class index : 0  name : Background
--- DataSetInfo              : Class index : 1  name : Signal
--- Factory                  : Booking method: REP_Estimator
--- REP_Estimator            : Building Network
--- REP_Estimator            : Initializing weights
--- DataSetFactory           : Splitmode is: "RANDOM" the mixmode is: "SAMEASSPLITMODE"
--- DataSetFactory           : Create training and testing trees -- looping over class "Background" ...
--- DataSetFactory           : Weight expression for class 'Background': "weight"
--- DataSetFactory           : Create training and testing trees -- looping over class "Signal" ...
--- DataSetFactory           : Weight expression for class 'Signal': "weight"
--- DataSetFactory           : Number of events in input trees (after possible flattening of arrays):
--- DataSetFactory           :     Background      -- number of events       : 90872  / sum of weights: 90872
--- DataSetFactory           :     Signal          -- number of events       : 173550  / sum of weights: 173550
--- DataSetFactory           :     Background tree -- total number of entries: 90872
--- DataSetFactory           :     Signal     tree -- total number of entries: 173550
--- DataSetFactory           : Preselection: (will NOT affect number of requested training and testing events)
--- DataSetFactory           :     Background requirement: "1"
--- DataSetFactory           :     Background      -- number of events passed: 90872  / sum of weights: 90872
--- DataSetFactory           :     Background      -- efficiency             : 1     
--- DataSetFactory           :     Signal     requirement: "1"
--- DataSetFactory           :     Signal          -- number of events passed: 173550  / sum of weights: 173550
--- DataSetFactory           :     Signal          -- efficiency             : 1     
--- DataSetFactory           : Weight renormalisation mode: "EqualNumEvents": renormalises all event classes ...
--- DataSetFactory           :  such that the effective (weighted) number of events in each class is the same 
--- DataSetFactory           :  (and equals the number of events (entries) given for class=0 )
--- DataSetFactory           : ... i.e. such that Sum[i=1..N_j]{w_i} = N_classA, j=classA, classB, ...
--- DataSetFactory           : ... (note that N_j is the sum of TRAINING events
--- DataSetFactory           :  ..... Testing events are not renormalised nor included in the renormalisation factor!)
--- DataSetFactory           : --> Rescale Background event weights by factor: 1
--- DataSetFactory           : --> Rescale Signal     event weights by factor: 0.523607
--- DataSetFactory           : Number of training and testing events after rescaling:
--- DataSetFactory           : ------------------------------------------------------
--- DataSetFactory           : Background -- training events            : 45436 (sum of weights: 45436) - requested were 0 events
--- DataSetFactory           : Background -- testing events             : 45436 (sum of weights: 45436) - requested were 0 events
--- DataSetFactory           : Background -- training and testing events: 90872 (sum of weights: 90872)
--- DataSetFactory           : Signal     -- training events            : 86775 (sum of weights: 45436) - requested were 0 events
--- DataSetFactory           : Signal     -- testing events             : 86775 (sum of weights: 86775) - requested were 0 events
--- DataSetFactory           : Signal     -- training and testing events: 173550 (sum of weights: 132211)
--- DataSetFactory           : Create internal training tree
--- DataSetFactory           : Create internal testing tree
--- DataSetInfo              : Correlation matrix (Background):
--- DataSetInfo              : -------------------------------------------------------------------------------------------------------------------
--- DataSetInfo              :                mult log_partPt log_partP  nnkrec log_ptB log_IPs partlcs  PIDNNk PIDNNpi  PIDNNp ghostProb log_IPPU
--- DataSetInfo              :       mult:  +1.000     +0.032    +0.044  +0.093  +0.068  -0.039  +0.093  -0.072  +0.038  +0.008    +0.030   -0.014
--- DataSetInfo              : log_partPt:  +0.032     +1.000    +0.494  -0.008  +0.058  +0.107  -0.025  +0.175  +0.017  -0.295    -0.059   +0.074
--- DataSetInfo              :  log_partP:  +0.044     +0.494    +1.000  +0.021  +0.030  -0.016  +0.057  +0.340  +0.112  -0.580    -0.162   -0.172
--- DataSetInfo              :     nnkrec:  +0.093     -0.008    +0.021  +1.000  +0.013  +0.022  +0.068  -0.093  +0.055  +0.028    +0.019   -0.705
--- DataSetInfo              :    log_ptB:  +0.068     +0.058    +0.030  +0.013  +1.000  +0.007  +0.017  +0.004  +0.003  -0.017    -0.003   -0.010
--- DataSetInfo              :    log_IPs:  -0.039     +0.107    -0.016  +0.022  +0.007  +1.000  -0.055  +0.022  -0.029  +0.022    -0.060   +0.001
--- DataSetInfo              :    partlcs:  +0.093     -0.025    +0.057  +0.068  +0.017  -0.055  +1.000  -0.135  -0.043  -0.093    +0.342   -0.066
--- DataSetInfo              :     PIDNNk:  -0.072     +0.175    +0.340  -0.093  +0.004  +0.022  -0.135  +1.000  -0.577  -0.403    -0.383   +0.032
--- DataSetInfo              :    PIDNNpi:  +0.038     +0.017    +0.112  +0.055  +0.003  -0.029  -0.043  -0.577  +1.000  -0.185    +0.279   -0.076
--- DataSetInfo              :     PIDNNp:  +0.008     -0.295    -0.580  +0.028  -0.017  +0.022  -0.093  -0.403  -0.185  +1.000    -0.089   +0.058
--- DataSetInfo              :  ghostProb:  +0.030     -0.059    -0.162  +0.019  -0.003  -0.060  +0.342  -0.383  +0.279  -0.089    +1.000   -0.001
--- DataSetInfo              :   log_IPPU:  -0.014     +0.074    -0.172  -0.705  -0.010  +0.001  -0.066  +0.032  -0.076  +0.058    -0.001   +1.000
--- DataSetInfo              : -------------------------------------------------------------------------------------------------------------------
--- DataSetInfo              : Correlation matrix (Signal):
--- DataSetInfo              : -------------------------------------------------------------------------------------------------------------------
--- DataSetInfo              :                mult log_partPt log_partP  nnkrec log_ptB log_IPs partlcs  PIDNNk PIDNNpi  PIDNNp ghostProb log_IPPU
--- DataSetInfo              :       mult:  +1.000     +0.031    +0.043  +0.099  +0.063  +0.005  +0.083  -0.092  +0.061  +0.024    +0.035   -0.017
--- DataSetInfo              : log_partPt:  +0.031     +1.000    +0.509  +0.000  +0.061  +0.113  -0.030  +0.199  +0.027  -0.323    -0.078   +0.071
--- DataSetInfo              :  log_partP:  +0.043     +0.509    +1.000  +0.003  +0.036  -0.028  +0.037  +0.368  +0.136  -0.610    -0.175   -0.140
--- DataSetInfo              :     nnkrec:  +0.099     +0.000    +0.003  +1.000  +0.008  +0.015  +0.070  -0.108  +0.062  +0.048    +0.028   -0.709
--- DataSetInfo              :    log_ptB:  +0.063     +0.061    +0.036  +0.008  +1.000  +0.009  +0.012  +0.004  +0.008  -0.024    -0.002   +0.001
--- DataSetInfo              :    log_IPs:  +0.005     +0.113    -0.028  +0.015  +0.009  +1.000  -0.037  +0.015  -0.035  +0.028    -0.044   +0.016
--- DataSetInfo              :    partlcs:  +0.083     -0.030    +0.037  +0.070  +0.012  -0.037  +1.000  -0.141  -0.029  -0.062    +0.342   -0.061
--- DataSetInfo              :     PIDNNk:  -0.092     +0.199    +0.368  -0.108  +0.004  +0.015  -0.141  +1.000  -0.566  -0.486    -0.389   +0.044
--- DataSetInfo              :    PIDNNpi:  +0.061     +0.027    +0.136  +0.062  +0.008  -0.035  -0.029  -0.566  +1.000  -0.123    +0.284   -0.082
--- DataSetInfo              :     PIDNNp:  +0.024     -0.323    -0.610  +0.048  -0.024  +0.028  -0.062  -0.486  -0.123  +1.000    -0.031   +0.040
--- DataSetInfo              :  ghostProb:  +0.035     -0.078    -0.175  +0.028  -0.002  -0.044  +0.342  -0.389  +0.284  -0.031    +1.000   -0.010
--- DataSetInfo              :   log_IPPU:  -0.017     +0.071    -0.140  -0.709  +0.001  +0.016  -0.061  +0.044  -0.082  +0.040    -0.010   +1.000
--- DataSetInfo              : -------------------------------------------------------------------------------------------------------------------
--- DataSetFactory           :  
--- Factory                  : 
--- Factory                  : current transformation string: 'I,D,N'
--- Factory                  : Create Transformation "I" with events from all classes.
--- Id                       : Transformation, Variable selection : 
--- Id                       : Input : variable 'mult' (index=0).   <---> Output : variable 'mult' (index=0).
--- Id                       : Input : variable 'log_partPt' (index=1).   <---> Output : variable 'log_partPt' (index=1).
--- Id                       : Input : variable 'log_partP' (index=2).   <---> Output : variable 'log_partP' (index=2).
--- Id                       : Input : variable 'nnkrec' (index=3).   <---> Output : variable 'nnkrec' (index=3).
--- Id                       : Input : variable 'log_ptB' (index=4).   <---> Output : variable 'log_ptB' (index=4).
--- Id                       : Input : variable 'log_IPs' (index=5).   <---> Output : variable 'log_IPs' (index=5).
--- Id                       : Input : variable 'partlcs' (index=6).   <---> Output : variable 'partlcs' (index=6).
--- Id                       : Input : variable 'PIDNNk' (index=7).   <---> Output : variable 'PIDNNk' (index=7).
--- Id                       : Input : variable 'PIDNNpi' (index=8).   <---> Output : variable 'PIDNNpi' (index=8).
--- Id                       : Input : variable 'PIDNNp' (index=9).   <---> Output : variable 'PIDNNp' (index=9).
--- Id                       : Input : variable 'ghostProb' (index=10).   <---> Output : variable 'ghostProb' (index=10).
--- Id                       : Input : variable 'log_IPPU' (index=11).   <---> Output : variable 'log_IPPU' (index=11).
--- Factory                  : Create Transformation "D" with events from all classes.
--- Deco                     : Transformation, Variable selection : 
--- Deco                     : Input : variable 'mult' (index=0).   <---> Output : variable 'mult' (index=0).
--- Deco                     : Input : variable 'log_partPt' (index=1).   <---> Output : variable 'log_partPt' (index=1).
--- Deco                     : Input : variable 'log_partP' (index=2).   <---> Output : variable 'log_partP' (index=2).
--- Deco                     : Input : variable 'nnkrec' (index=3).   <---> Output : variable 'nnkrec' (index=3).
--- Deco                     : Input : variable 'log_ptB' (index=4).   <---> Output : variable 'log_ptB' (index=4).
--- Deco                     : Input : variable 'log_IPs' (index=5).   <---> Output : variable 'log_IPs' (index=5).
--- Deco                     : Input : variable 'partlcs' (index=6).   <---> Output : variable 'partlcs' (index=6).
--- Deco                     : Input : variable 'PIDNNk' (index=7).   <---> Output : variable 'PIDNNk' (index=7).
--- Deco                     : Input : variable 'PIDNNpi' (index=8).   <---> Output : variable 'PIDNNpi' (index=8).
--- Deco                     : Input : variable 'PIDNNp' (index=9).   <---> Output : variable 'PIDNNp' (index=9).
--- Deco                     : Input : variable 'ghostProb' (index=10).   <---> Output : variable 'ghostProb' (index=10).
--- Deco                     : Input : variable 'log_IPPU' (index=11).   <---> Output : variable 'log_IPPU' (index=11).
--- Factory                  : Create Transformation "N" with events from all classes.
--- Norm                     : Transformation, Variable selection : 
--- Norm                     : Input : variable 'mult' (index=0).   <---> Output : variable 'mult' (index=0).
--- Norm                     : Input : variable 'log_partPt' (index=1).   <---> Output : variable 'log_partPt' (index=1).
--- Norm                     : Input : variable 'log_partP' (index=2).   <---> Output : variable 'log_partP' (index=2).
--- Norm                     : Input : variable 'nnkrec' (index=3).   <---> Output : variable 'nnkrec' (index=3).
--- Norm                     : Input : variable 'log_ptB' (index=4).   <---> Output : variable 'log_ptB' (index=4).
--- Norm                     : Input : variable 'log_IPs' (index=5).   <---> Output : variable 'log_IPs' (index=5).
--- Norm                     : Input : variable 'partlcs' (index=6).   <---> Output : variable 'partlcs' (index=6).
--- Norm                     : Input : variable 'PIDNNk' (index=7).   <---> Output : variable 'PIDNNk' (index=7).
--- Norm                     : Input : variable 'PIDNNpi' (index=8).   <---> Output : variable 'PIDNNpi' (index=8).
--- Norm                     : Input : variable 'PIDNNp' (index=9).   <---> Output : variable 'PIDNNp' (index=9).
--- Norm                     : Input : variable 'ghostProb' (index=10).   <---> Output : variable 'ghostProb' (index=10).
--- Norm                     : Input : variable 'log_IPPU' (index=11).   <---> Output : variable 'log_IPPU' (index=11).
--- Id                       : Preparing the Identity transformation...
--- Deco                     : Preparing the Decorrelation transformation...
--- Norm                     : Preparing the transformation.
--- TFHandler_Factory        : ---------------------------------------------------------------------
--- TFHandler_Factory        :   Variable          Mean          RMS   [        Min          Max ]
--- TFHandler_Factory        : ---------------------------------------------------------------------
--- TFHandler_Factory        :       mult:    -0.56198     0.20221   [     -1.0000      1.0000 ]
--- TFHandler_Factory        : log_partPt:    -0.31738     0.28009   [     -1.0000      1.0000 ]
--- TFHandler_Factory        :  log_partP:    -0.14359     0.31526   [     -1.0000      1.0000 ]
--- TFHandler_Factory        :     nnkrec:    -0.59691     0.20694   [     -1.0000      1.0000 ]
--- TFHandler_Factory        :    log_ptB:     0.41020     0.17848   [     -1.0000      1.0000 ]
--- TFHandler_Factory        :    log_IPs:    -0.36981     0.37993   [     -1.0000      1.0000 ]
--- TFHandler_Factory        :    partlcs:    -0.25539     0.22532   [     -1.0000      1.0000 ]
--- TFHandler_Factory        :     PIDNNk:     0.20115     0.25907   [     -1.0000      1.0000 ]
--- TFHandler_Factory        :    PIDNNpi:    -0.20809     0.24473   [     -1.0000      1.0000 ]
--- TFHandler_Factory        :     PIDNNp:    -0.19345     0.24278   [     -1.0000      1.0000 ]
--- TFHandler_Factory        :  ghostProb:    -0.49518     0.18502   [     -1.0000      1.0000 ]
--- TFHandler_Factory        :   log_IPPU:    0.063988     0.45325   [     -1.0000      1.0000 ]
--- TFHandler_Factory        : ---------------------------------------------------------------------
--- TFHandler_Factory        : Plot event variables for Id_Deco_Norm
--- TFHandler_Factory        : Create scatter and profile plots in target-file directory: 
--- TFHandler_Factory        : /mnt/mfs/notebook/analyses/tagging_LHCb/MC/tmpRGTbeB/result.root:/InputVariables_Id_Deco_Norm/CorrelationPlots
--- TFHandler_Factory        :  
--- TFHandler_Factory        : Ranking input variables (method unspecific)...
--- Id_Deco_NormTransforma...: Ranking result (top variable is best ranked)
--- Id_Deco_NormTransforma...: -----------------------------------
--- Id_Deco_NormTransforma...: Rank : Variable   : Separation
--- Id_Deco_NormTransforma...: -----------------------------------
--- Id_Deco_NormTransforma...:    1 : PIDNNk     : 6.876e-03
--- Id_Deco_NormTransforma...:    2 : ghostProb  : 4.069e-03
--- Id_Deco_NormTransforma...:    3 : log_partP  : 3.320e-03
--- Id_Deco_NormTransforma...:    4 : PIDNNpi    : 3.225e-03
--- Id_Deco_NormTransforma...:    5 : log_IPs    : 3.215e-03
--- Id_Deco_NormTransforma...:    6 : log_partPt : 2.866e-03
--- Id_Deco_NormTransforma...:    7 : PIDNNp     : 2.578e-03
--- Id_Deco_NormTransforma...:    8 : mult       : 2.200e-03
--- Id_Deco_NormTransforma...:    9 : partlcs    : 1.469e-03
--- Id_Deco_NormTransforma...:   10 : log_IPPU   : 1.451e-03
--- Id_Deco_NormTransforma...:   11 : nnkrec     : 1.199e-03
--- Id_Deco_NormTransforma...:   12 : log_ptB    : 3.238e-04
--- Id_Deco_NormTransforma...: -----------------------------------
--- Factory                  :  
--- Factory                  : Train all methods for Classification ...
--- Factory                  : Train method: REP_Estimator for Classification
--- REP_Estimator            : Begin training
--- REP_Estimator            : Training Network
Error in <TDecompLU::InvertLU>: matrix is singular, 20 diag elements < tolerance of 2.2204e-16
--- REP_Estimator            : Finalizing handling of Regulator terms, trainE=1.37896 testE=1.38109
--- REP_Estimator            : Done with handling of Regulator terms
--- REP_Estimator            : End of training                                              
--- REP_Estimator            : Elapsed time for training with 132211 events: 1.37e+03 sec         
--- REP_Estimator            : Create MVA output for classification on training sample
--- REP_Estimator            : Evaluation of REP_Estimator on training sample (132211 events)
--- REP_Estimator            : Elapsed time for evaluation of 132211 events: 0.407 sec       
--- REP_Estimator            : Creating weight file in xml format: weights/TMVAEstimation_REP_Estimator.weights.xml
--- REP_Estimator            : Creating standalone response class: weights/TMVAEstimation_REP_Estimator.class.C
--- REP_Estimator            : Write special histos to file: /mnt/mfs/notebook/analyses/tagging_LHCb/MC/tmpRGTbeB/result.root:/Method_MLP/REP_Estimator
--- Factory                  : Training finished
--- Factory                  : 
--- Factory                  : Ranking input variables (method specific)...
--- REP_Estimator            : Ranking result (top variable is best ranked)
--- REP_Estimator            : -----------------------------------
--- REP_Estimator            : Rank : Variable   : Importance
--- REP_Estimator            : -----------------------------------
--- REP_Estimator            :    1 : mult       : 2.858e+04
--- REP_Estimator            :    2 : log_IPPU   : 6.341e+02
--- REP_Estimator            :    3 : log_partP  : 1.912e+02
--- REP_Estimator            :    4 : log_IPs    : 1.606e+02
--- REP_Estimator            :    5 : nnkrec     : 1.068e+02
--- REP_Estimator            :    6 : log_ptB    : 4.286e+01
--- REP_Estimator            :    7 : partlcs    : 2.281e+01
--- REP_Estimator            :    8 : PIDNNk     : 9.110e+00
--- REP_Estimator            :    9 : log_partPt : 5.055e+00
--- REP_Estimator            :   10 : PIDNNpi    : 1.075e+00
--- REP_Estimator            :   11 : PIDNNp     : 3.461e-01
--- REP_Estimator            :   12 : ghostProb  : 1.316e-01
--- REP_Estimator            : -----------------------------------
--- Factory                  : 
--- Factory                  : === Destroy and recreate all methods via weight files for testing ===
--- Factory                  : 
--- MethodBase               : Reading weight file: weights/TMVAEstimation_REP_Estimator.weights.xml
--- REP_Estimator            : Read method "REP_Estimator" of type "MLP"
--- REP_Estimator            : MVA method was trained with TMVA Version: 4.2.0
--- REP_Estimator            : MVA method was trained with ROOT Version: 5.34/32
--- REP_Estimator            : Building Network
--- REP_Estimator            : Initializing weights

#error "You need a ISO C conforming compiler to use the glibc headers"
*** Interpreter error recovered ***
--- Factory                  : You are running ROOT Version: 5.34/32, Jun 23, 2015
--- Factory                  : 
--- Factory                  : _/_/_/_/_/ _|      _|  _|      _|    _|_|   
--- Factory                  :    _/      _|_|  _|_|  _|      _|  _|    _| 
--- Factory                  :   _/       _|  _|  _|  _|      _|  _|_|_|_| 
--- Factory                  :  _/        _|      _|    _|  _|    _|    _| 
--- Factory                  : _/         _|      _|      _|      _|    _| 
--- Factory                  : 
--- Factory                  : ___________TMVA Version 4.2.0, Sep 19, 2013
--- Factory                  : 
--- DataSetInfo              : Added class "Background"	 with internal class number 0
--- DataSetInfo              : Added class "Signal"	 with internal class number 1
--- Factory                  : Add Tree TrainAssignTree_Background of type Background with 45392 events
--- Factory                  : Add Tree TestAssignTree_Background of type Background with 45392 events
--- Factory                  : Add Tree TrainAssignTree_Signal of type Signal with 86819 events
--- Factory                  : Add Tree TestAssignTree_Signal of type Signal with 86819 events
--- DataSetInfo              : Class index : 0  name : Background
--- DataSetInfo              : Class index : 1  name : Signal
--- Factory                  : Booking method: REP_Estimator
--- REP_Estimator            : Building Network
--- REP_Estimator            : Initializing weights
--- DataSetFactory           : Splitmode is: "RANDOM" the mixmode is: "SAMEASSPLITMODE"
--- DataSetFactory           : Create training and testing trees -- looping over class "Background" ...
--- DataSetFactory           : Weight expression for class 'Background': "weight"
--- DataSetFactory           : Create training and testing trees -- looping over class "Signal" ...
--- DataSetFactory           : Weight expression for class 'Signal': "weight"
--- DataSetFactory           : Number of events in input trees (after possible flattening of arrays):
--- DataSetFactory           :     Background      -- number of events       : 90784  / sum of weights: 90784
--- DataSetFactory           :     Signal          -- number of events       : 173638  / sum of weights: 173638
--- DataSetFactory           :     Background tree -- total number of entries: 90784
--- DataSetFactory           :     Signal     tree -- total number of entries: 173638
--- DataSetFactory           : Preselection: (will NOT affect number of requested training and testing events)
--- DataSetFactory           :     Background requirement: "1"
--- DataSetFactory           :     Background      -- number of events passed: 90784  / sum of weights: 90784
--- DataSetFactory           :     Background      -- efficiency             : 1     
--- DataSetFactory           :     Signal     requirement: "1"
--- DataSetFactory           :     Signal          -- number of events passed: 173638  / sum of weights: 173638
--- DataSetFactory           :     Signal          -- efficiency             : 1     
--- DataSetFactory           : Weight renormalisation mode: "EqualNumEvents": renormalises all event classes ...
--- DataSetFactory           :  such that the effective (weighted) number of events in each class is the same 
--- DataSetFactory           :  (and equals the number of events (entries) given for class=0 )
--- DataSetFactory           : ... i.e. such that Sum[i=1..N_j]{w_i} = N_classA, j=classA, classB, ...
--- DataSetFactory           : ... (note that N_j is the sum of TRAINING events
--- DataSetFactory           :  ..... Testing events are not renormalised nor included in the renormalisation factor!)
--- DataSetFactory           : --> Rescale Background event weights by factor: 1
--- DataSetFactory           : --> Rescale Signal     event weights by factor: 0.522835
--- DataSetFactory           : Number of training and testing events after rescaling:
--- DataSetFactory           : ------------------------------------------------------
--- DataSetFactory           : Background -- training events            : 45392 (sum of weights: 45392) - requested were 0 events
--- DataSetFactory           : Background -- testing events             : 45392 (sum of weights: 45392) - requested were 0 events
--- DataSetFactory           : Background -- training and testing events: 90784 (sum of weights: 90784)
--- DataSetFactory           : Signal     -- training events            : 86819 (sum of weights: 45392) - requested were 0 events
--- DataSetFactory           : Signal     -- testing events             : 86819 (sum of weights: 86819) - requested were 0 events
--- DataSetFactory           : Signal     -- training and testing events: 173638 (sum of weights: 132211)
--- DataSetFactory           : Create internal training tree
--- DataSetFactory           : Create internal testing tree
--- DataSetInfo              : Correlation matrix (Background):
--- DataSetInfo              : -------------------------------------------------------------------------------------------------------------------
--- DataSetInfo              :                mult log_partPt log_partP  nnkrec log_ptB log_IPs partlcs  PIDNNk PIDNNpi  PIDNNp ghostProb log_IPPU
--- DataSetInfo              :       mult:  +1.000     +0.030    +0.042  +0.087  +0.070  -0.039  +0.093  -0.078  +0.047  +0.005    +0.033   -0.003
--- DataSetInfo              : log_partPt:  +0.030     +1.000    +0.502  +0.006  +0.081  +0.110  -0.024  +0.164  +0.025  -0.296    -0.068   +0.064
--- DataSetInfo              :  log_partP:  +0.042     +0.502    +1.000  +0.021  +0.052  -0.004  +0.043  +0.335  +0.118  -0.586    -0.170   -0.170
--- DataSetInfo              :     nnkrec:  +0.087     +0.006    +0.021  +1.000  +0.007  +0.018  +0.071  -0.102  +0.059  +0.030    +0.025   -0.701
--- DataSetInfo              :    log_ptB:  +0.070     +0.081    +0.052  +0.007  +1.000  +0.006  +0.016  +0.003  +0.006  -0.034    -0.000   -0.000
--- DataSetInfo              :    log_IPs:  -0.039     +0.110    -0.004  +0.018  +0.006  +1.000  -0.057  +0.019  -0.022  +0.029    -0.063   -0.001
--- DataSetInfo              :    partlcs:  +0.093     -0.024    +0.043  +0.071  +0.016  -0.057  +1.000  -0.138  -0.042  -0.084    +0.354   -0.067
--- DataSetInfo              :     PIDNNk:  -0.078     +0.164    +0.335  -0.102  +0.003  +0.019  -0.138  +1.000  -0.581  -0.401    -0.378   +0.038
--- DataSetInfo              :    PIDNNpi:  +0.047     +0.025    +0.118  +0.059  +0.006  -0.022  -0.042  -0.581  +1.000  -0.186    +0.276   -0.073
--- DataSetInfo              :     PIDNNp:  +0.005     -0.296    -0.586  +0.030  -0.034  +0.029  -0.084  -0.401  -0.186  +1.000    -0.084   +0.058
--- DataSetInfo              :  ghostProb:  +0.033     -0.068    -0.170  +0.025  -0.000  -0.063  +0.354  -0.378  +0.276  -0.084    +1.000   -0.004
--- DataSetInfo              :   log_IPPU:  -0.003     +0.064    -0.170  -0.701  -0.000  -0.001  -0.067  +0.038  -0.073  +0.058    -0.004   +1.000
--- DataSetInfo              : -------------------------------------------------------------------------------------------------------------------
--- DataSetInfo              : Correlation matrix (Signal):
--- DataSetInfo              : -------------------------------------------------------------------------------------------------------------------
--- DataSetInfo              :                mult log_partPt log_partP  nnkrec log_ptB log_IPs partlcs  PIDNNk PIDNNpi  PIDNNp ghostProb log_IPPU
--- DataSetInfo              :       mult:  +1.000     +0.029    +0.049  +0.096  +0.067  -0.001  +0.087  -0.082  +0.052  +0.016    +0.029   -0.018
--- DataSetInfo              : log_partPt:  +0.029     +1.000    +0.506  +0.009  +0.061  +0.117  -0.029  +0.208  +0.023  -0.323    -0.075   +0.065
--- DataSetInfo              :  log_partP:  +0.049     +0.506    +1.000  +0.012  +0.035  -0.025  +0.039  +0.372  +0.133  -0.611    -0.177   -0.151
--- DataSetInfo              :     nnkrec:  +0.096     +0.009    +0.012  +1.000  +0.007  +0.013  +0.071  -0.103  +0.057  +0.041    +0.026   -0.709
--- DataSetInfo              :    log_ptB:  +0.067     +0.061    +0.035  +0.007  +1.000  +0.009  +0.008  -0.001  +0.010  -0.019    -0.001   +0.004
--- DataSetInfo              :    log_IPs:  -0.001     +0.117    -0.025  +0.013  +0.009  +1.000  -0.036  +0.018  -0.029  +0.023    -0.046   +0.018
--- DataSetInfo              :    partlcs:  +0.087     -0.029    +0.039  +0.071  +0.008  -0.036  +1.000  -0.136  -0.036  -0.064    +0.328   -0.067
--- DataSetInfo              :     PIDNNk:  -0.082     +0.208    +0.372  -0.103  -0.001  +0.018  -0.136  +1.000  -0.562  -0.487    -0.389   +0.042
--- DataSetInfo              :    PIDNNpi:  +0.052     +0.023    +0.133  +0.057  +0.010  -0.029  -0.036  -0.562  +1.000  -0.123    +0.281   -0.077
--- DataSetInfo              :     PIDNNp:  +0.016     -0.323    -0.611  +0.041  -0.019  +0.023  -0.064  -0.487  -0.123  +1.000    -0.030   +0.045
--- DataSetInfo              :  ghostProb:  +0.029     -0.075    -0.177  +0.026  -0.001  -0.046  +0.328  -0.389  +0.281  -0.030    +1.000   -0.011
--- DataSetInfo              :   log_IPPU:  -0.018     +0.065    -0.151  -0.709  +0.004  +0.018  -0.067  +0.042  -0.077  +0.045    -0.011   +1.000
--- DataSetInfo              : -------------------------------------------------------------------------------------------------------------------
--- DataSetFactory           :  
--- Factory                  : 
--- Factory                  : current transformation string: 'I,D,N'
--- Factory                  : Create Transformation "I" with events from all classes.
--- Id                       : Transformation, Variable selection : 
--- Id                       : Input : variable 'mult' (index=0).   <---> Output : variable 'mult' (index=0).
--- Id                       : Input : variable 'log_partPt' (index=1).   <---> Output : variable 'log_partPt' (index=1).
--- Id                       : Input : variable 'log_partP' (index=2).   <---> Output : variable 'log_partP' (index=2).
--- Id                       : Input : variable 'nnkrec' (index=3).   <---> Output : variable 'nnkrec' (index=3).
--- Id                       : Input : variable 'log_ptB' (index=4).   <---> Output : variable 'log_ptB' (index=4).
--- Id                       : Input : variable 'log_IPs' (index=5).   <---> Output : variable 'log_IPs' (index=5).
--- Id                       : Input : variable 'partlcs' (index=6).   <---> Output : variable 'partlcs' (index=6).
--- Id                       : Input : variable 'PIDNNk' (index=7).   <---> Output : variable 'PIDNNk' (index=7).
--- Id                       : Input : variable 'PIDNNpi' (index=8).   <---> Output : variable 'PIDNNpi' (index=8).
--- Id                       : Input : variable 'PIDNNp' (index=9).   <---> Output : variable 'PIDNNp' (index=9).
--- Id                       : Input : variable 'ghostProb' (index=10).   <---> Output : variable 'ghostProb' (index=10).
--- Id                       : Input : variable 'log_IPPU' (index=11).   <---> Output : variable 'log_IPPU' (index=11).
--- Factory                  : Create Transformation "D" with events from all classes.
--- Deco                     : Transformation, Variable selection : 
--- Deco                     : Input : variable 'mult' (index=0).   <---> Output : variable 'mult' (index=0).
--- Deco                     : Input : variable 'log_partPt' (index=1).   <---> Output : variable 'log_partPt' (index=1).
--- Deco                     : Input : variable 'log_partP' (index=2).   <---> Output : variable 'log_partP' (index=2).
--- Deco                     : Input : variable 'nnkrec' (index=3).   <---> Output : variable 'nnkrec' (index=3).
--- Deco                     : Input : variable 'log_ptB' (index=4).   <---> Output : variable 'log_ptB' (index=4).
--- Deco                     : Input : variable 'log_IPs' (index=5).   <---> Output : variable 'log_IPs' (index=5).
--- Deco                     : Input : variable 'partlcs' (index=6).   <---> Output : variable 'partlcs' (index=6).
--- Deco                     : Input : variable 'PIDNNk' (index=7).   <---> Output : variable 'PIDNNk' (index=7).
--- Deco                     : Input : variable 'PIDNNpi' (index=8).   <---> Output : variable 'PIDNNpi' (index=8).
--- Deco                     : Input : variable 'PIDNNp' (index=9).   <---> Output : variable 'PIDNNp' (index=9).
--- Deco                     : Input : variable 'ghostProb' (index=10).   <---> Output : variable 'ghostProb' (index=10).
--- Deco                     : Input : variable 'log_IPPU' (index=11).   <---> Output : variable 'log_IPPU' (index=11).
--- Factory                  : Create Transformation "N" with events from all classes.
--- Norm                     : Transformation, Variable selection : 
--- Norm                     : Input : variable 'mult' (index=0).   <---> Output : variable 'mult' (index=0).
--- Norm                     : Input : variable 'log_partPt' (index=1).   <---> Output : variable 'log_partPt' (index=1).
--- Norm                     : Input : variable 'log_partP' (index=2).   <---> Output : variable 'log_partP' (index=2).
--- Norm                     : Input : variable 'nnkrec' (index=3).   <---> Output : variable 'nnkrec' (index=3).
--- Norm                     : Input : variable 'log_ptB' (index=4).   <---> Output : variable 'log_ptB' (index=4).
--- Norm                     : Input : variable 'log_IPs' (index=5).   <---> Output : variable 'log_IPs' (index=5).
--- Norm                     : Input : variable 'partlcs' (index=6).   <---> Output : variable 'partlcs' (index=6).
--- Norm                     : Input : variable 'PIDNNk' (index=7).   <---> Output : variable 'PIDNNk' (index=7).
--- Norm                     : Input : variable 'PIDNNpi' (index=8).   <---> Output : variable 'PIDNNpi' (index=8).
--- Norm                     : Input : variable 'PIDNNp' (index=9).   <---> Output : variable 'PIDNNp' (index=9).
--- Norm                     : Input : variable 'ghostProb' (index=10).   <---> Output : variable 'ghostProb' (index=10).
--- Norm                     : Input : variable 'log_IPPU' (index=11).   <---> Output : variable 'log_IPPU' (index=11).
--- Id                       : Preparing the Identity transformation...
--- Deco                     : Preparing the Decorrelation transformation...
--- Norm                     : Preparing the transformation.
--- TFHandler_Factory        : ---------------------------------------------------------------------
--- TFHandler_Factory        :   Variable          Mean          RMS   [        Min          Max ]
--- TFHandler_Factory        : ---------------------------------------------------------------------
--- TFHandler_Factory        :       mult:    -0.51598     0.22415   [     -1.0000      1.0000 ]
--- TFHandler_Factory        : log_partPt:    -0.27038     0.29437   [     -1.0000      1.0000 ]
--- TFHandler_Factory        :  log_partP:    -0.16354     0.31476   [     -1.0000      1.0000 ]
--- TFHandler_Factory        :     nnkrec:    -0.57206     0.21606   [     -1.0000      1.0000 ]
--- TFHandler_Factory        :    log_ptB:     0.34107     0.16899   [     -1.0000      1.0000 ]
--- TFHandler_Factory        :    log_IPs:    -0.33381     0.39984   [     -1.0000      1.0000 ]
--- TFHandler_Factory        :    partlcs:    -0.21200     0.21801   [     -1.0000      1.0000 ]
--- TFHandler_Factory        :     PIDNNk:     0.19051     0.25417   [     -1.0000      1.0000 ]
--- TFHandler_Factory        :    PIDNNpi:    -0.20999     0.24513   [     -1.0000      1.0000 ]
--- TFHandler_Factory        :     PIDNNp:    -0.32562     0.20829   [     -1.0000      1.0000 ]
--- TFHandler_Factory        :  ghostProb:    -0.49164     0.19077   [     -1.0000      1.0000 ]
--- TFHandler_Factory        :   log_IPPU:    0.085178     0.46772   [     -1.0000      1.0000 ]
--- TFHandler_Factory        : ---------------------------------------------------------------------
--- TFHandler_Factory        : Plot event variables for Id_Deco_Norm
--- TFHandler_Factory        : Create scatter and profile plots in target-file directory: 
--- TFHandler_Factory        : /mnt/mfs/notebook/analyses/tagging_LHCb/MC/tmpSK2A82/result.root:/InputVariables_Id_Deco_Norm/CorrelationPlots
--- TFHandler_Factory        :  
--- TFHandler_Factory        : Ranking input variables (method unspecific)...
--- Id_Deco_NormTransforma...: Ranking result (top variable is best ranked)
--- Id_Deco_NormTransforma...: -----------------------------------
--- Id_Deco_NormTransforma...: Rank : Variable   : Separation
--- Id_Deco_NormTransforma...: -----------------------------------
--- Id_Deco_NormTransforma...:    1 : PIDNNk     : 6.769e-03
--- Id_Deco_NormTransforma...:    2 : ghostProb  : 4.277e-03
--- Id_Deco_NormTransforma...:    3 : log_partP  : 3.752e-03
--- Id_Deco_NormTransforma...:    4 : log_IPs    : 3.363e-03
--- Id_Deco_NormTransforma...:    5 : PIDNNpi    : 3.118e-03
--- Id_Deco_NormTransforma...:    6 : log_partPt : 2.600e-03
--- Id_Deco_NormTransforma...:    7 : mult       : 2.364e-03
--- Id_Deco_NormTransforma...:    8 : PIDNNp     : 1.945e-03
--- Id_Deco_NormTransforma...:    9 : log_IPPU   : 1.439e-03
--- Id_Deco_NormTransforma...:   10 : nnkrec     : 1.274e-03
--- Id_Deco_NormTransforma...:   11 : partlcs    : 1.221e-03
--- Id_Deco_NormTransforma...:   12 : log_ptB    : 3.408e-04
--- Id_Deco_NormTransforma...: -----------------------------------
--- Factory                  :  
--- Factory                  : Train all methods for Classification ...
--- Factory                  : Train method: REP_Estimator for Classification
--- REP_Estimator            : Begin training
--- REP_Estimator            : Training Network
Error in <TDecompLU::InvertLU>: matrix is singular, 17 diag elements < tolerance of 2.2204e-16
Error in <TDecompLU::InvertLU>: matrix is singular, 1 diag elements < tolerance of 2.2204e-16
Error in <TDecompLU::InvertLU>: matrix is singular, 2 diag elements < tolerance of 2.2204e-16
Error in <TDecompLU::InvertLU>: matrix is singular, 1 diag elements < tolerance of 2.2204e-16
Error in <TDecompLU::InvertLU>: matrix is singular, 1 diag elements < tolerance of 2.2204e-16
--- REP_Estimator            : Finalizing handling of Regulator terms, trainE=1.37823 testE=1.37848
--- REP_Estimator            : Done with handling of Regulator terms
--- REP_Estimator            : End of training                                              
--- REP_Estimator            : Elapsed time for training with 132211 events: 1.35e+03 sec         
--- REP_Estimator            : Create MVA output for classification on training sample
--- REP_Estimator            : Evaluation of REP_Estimator on training sample (132211 events)
--- REP_Estimator            : Elapsed time for evaluation of 132211 events: 0.39 sec       
--- REP_Estimator            : Creating weight file in xml format: weights/TMVAEstimation_REP_Estimator.weights.xml
--- REP_Estimator            : Creating standalone response class: weights/TMVAEstimation_REP_Estimator.class.C
--- REP_Estimator            : Write special histos to file: /mnt/mfs/notebook/analyses/tagging_LHCb/MC/tmpSK2A82/result.root:/Method_MLP/REP_Estimator
--- Factory                  : Training finished
--- Factory                  : 
--- Factory                  : Ranking input variables (method specific)...
--- REP_Estimator            : Ranking result (top variable is best ranked)
--- REP_Estimator            : -----------------------------------
--- REP_Estimator            : Rank : Variable   : Importance
--- REP_Estimator            : -----------------------------------
--- REP_Estimator            :    1 : mult       : 2.925e+04
--- REP_Estimator            :    2 : log_IPPU   : 5.348e+02
--- REP_Estimator            :    3 : log_partP  : 2.194e+02
--- REP_Estimator            :    4 : log_IPs    : 1.543e+02
--- REP_Estimator            :    5 : nnkrec     : 1.046e+02
--- REP_Estimator            :    6 : log_ptB    : 4.242e+01
--- REP_Estimator            :    7 : partlcs    : 2.127e+01
--- REP_Estimator            :    8 : PIDNNk     : 9.335e+00
--- REP_Estimator            :    9 : log_partPt : 5.213e+00
--- REP_Estimator            :   10 : PIDNNpi    : 1.035e+00
--- REP_Estimator            :   11 : PIDNNp     : 3.427e-01
--- REP_Estimator            :   12 : ghostProb  : 1.271e-01
--- REP_Estimator            : -----------------------------------
--- Factory                  : 
--- Factory                  : === Destroy and recreate all methods via weight files for testing ===
--- Factory                  : 
--- MethodBase               : Reading weight file: weights/TMVAEstimation_REP_Estimator.weights.xml
--- REP_Estimator            : Read method "REP_Estimator" of type "MLP"
--- REP_Estimator            : MVA method was trained with TMVA Version: 4.2.0
--- REP_Estimator            : MVA method was trained with ROOT Version: 5.34/32
--- REP_Estimator            : Building Network
--- REP_Estimator            : Initializing weights

#error "You need a ISO C conforming compiler to use the glibc headers"
*** Interpreter error recovered ***
--- Factory                  : You are running ROOT Version: 5.34/32, Jun 23, 2015
--- Factory                  : 
--- Factory                  : _/_/_/_/_/ _|      _|  _|      _|    _|_|   
--- Factory                  :    _/      _|_|  _|_|  _|      _|  _|    _| 
--- Factory                  :   _/       _|  _|  _|  _|      _|  _|_|_|_| 
--- Factory                  :  _/        _|      _|    _|  _|    _|    _| 
--- Factory                  : _/         _|      _|      _|      _|    _| 
--- Factory                  : 
--- Factory                  : ___________TMVA Version 4.2.0, Sep 19, 2013
--- Factory                  : 
--- DataSetInfo              : Added class "Signal"	 with internal class number 0
--- DataSetInfo              : Added class "Background"	 with internal class number 1
--- Factory                  : Add Tree TrainAssignTree_Signal of type Signal with 10037 events
--- Factory                  : Add Tree TestAssignTree_Signal of type Signal with 10037 events
--- Factory                  : Add Tree TrainAssignTree_Background of type Background with 5023 events
--- Factory                  : Add Tree TestAssignTree_Background of type Background with 5023 events
--- DataSetInfo              : Class index : 0  name : Signal
--- DataSetInfo              : Class index : 1  name : Background
--- Factory                  : Booking method: REP_Estimator
--- REP_Estimator            : Building Network
--- REP_Estimator            : Initializing weights
--- DataSetFactory           : Splitmode is: "RANDOM" the mixmode is: "SAMEASSPLITMODE"
--- DataSetFactory           : Create training and testing trees -- looping over class "Signal" ...
--- DataSetFactory           : Weight expression for class 'Signal': "weight"
--- DataSetFactory           : Create training and testing trees -- looping over class "Background" ...
--- DataSetFactory           : Weight expression for class 'Background': "weight"
--- DataSetFactory           : Number of events in input trees (after possible flattening of arrays):
--- DataSetFactory           :     Signal          -- number of events       : 20074  / sum of weights: 20074
--- DataSetFactory           :     Background      -- number of events       : 10046  / sum of weights: 10046
--- DataSetFactory           :     Signal     tree -- total number of entries: 20074
--- DataSetFactory           :     Background tree -- total number of entries: 10046
--- DataSetFactory           : Preselection: (will NOT affect number of requested training and testing events)
--- DataSetFactory           :     Signal     requirement: "1"
--- DataSetFactory           :     Signal          -- number of events passed: 20074  / sum of weights: 20074
--- DataSetFactory           :     Signal          -- efficiency             : 1     
--- DataSetFactory           :     Background requirement: "1"
--- DataSetFactory           :     Background      -- number of events passed: 10046  / sum of weights: 10046
--- DataSetFactory           :     Background      -- efficiency             : 1     
--- DataSetFactory           : Weight renormalisation mode: "EqualNumEvents": renormalises all event classes ...
--- DataSetFactory           :  such that the effective (weighted) number of events in each class is the same 
--- DataSetFactory           :  (and equals the number of events (entries) given for class=0 )
--- DataSetFactory           : ... i.e. such that Sum[i=1..N_j]{w_i} = N_classA, j=classA, classB, ...
--- DataSetFactory           : ... (note that N_j is the sum of TRAINING events
--- DataSetFactory           :  ..... Testing events are not renormalised nor included in the renormalisation factor!)
--- DataSetFactory           : --> Rescale Signal     event weights by factor: 1
--- DataSetFactory           : --> Rescale Background event weights by factor: 1.99821
--- DataSetFactory           : Number of training and testing events after rescaling:
--- DataSetFactory           : ------------------------------------------------------
--- DataSetFactory           : Signal     -- training events            : 10037 (sum of weights: 10037) - requested were 0 events
--- DataSetFactory           : Signal     -- testing events             : 10037 (sum of weights: 10037) - requested were 0 events
--- DataSetFactory           : Signal     -- training and testing events: 20074 (sum of weights: 20074)
--- DataSetFactory           : Background -- training events            : 5023 (sum of weights: 10037) - requested were 0 events
--- DataSetFactory           : Background -- testing events             : 5023 (sum of weights: 5023) - requested were 0 events
--- DataSetFactory           : Background -- training and testing events: 10046 (sum of weights: 15060)
--- DataSetFactory           : Create internal training tree
--- DataSetFactory           : Create internal testing tree
--- DataSetInfo              : Correlation matrix (Signal):
--- DataSetInfo              : ----------------------------------------------------------------------------------------------
--- DataSetInfo              :                mult log_partPt log_partP log_ptB log_IPs partlcs log_eOverP ghostProb log_IPPU
--- DataSetInfo              :       mult:  +1.000     +0.019    +0.035  +0.070  +0.062  +0.105     +0.080    +0.121   -0.054
--- DataSetInfo              : log_partPt:  +0.019     +1.000    +0.559  +0.048  -0.029  -0.068     -0.041    -0.079   +0.092
--- DataSetInfo              :  log_partP:  +0.035     +0.559    +1.000  +0.027  -0.123  +0.053     -0.025    +0.129   -0.117
--- DataSetInfo              :    log_ptB:  +0.070     +0.048    +0.027  +1.000  +0.009  +0.009     +0.014    +0.017   +0.000
--- DataSetInfo              :    log_IPs:  +0.062     -0.029    -0.123  +0.009  +1.000  -0.007     +0.005    -0.010   -0.026
--- DataSetInfo              :    partlcs:  +0.105     -0.068    +0.053  +0.009  -0.007  +1.000     +0.014    +0.738   -0.104
--- DataSetInfo              : log_eOverP:  +0.080     -0.041    -0.025  +0.014  +0.005  +0.014     +1.000    +0.020   -0.066
--- DataSetInfo              :  ghostProb:  +0.121     -0.079    +0.129  +0.017  -0.010  +0.738     +0.020    +1.000   -0.147
--- DataSetInfo              :   log_IPPU:  -0.054     +0.092    -0.117  +0.000  -0.026  -0.104     -0.066    -0.147   +1.000
--- DataSetInfo              : ----------------------------------------------------------------------------------------------
--- DataSetInfo              : Correlation matrix (Background):
--- DataSetInfo              : ----------------------------------------------------------------------------------------------
--- DataSetInfo              :                mult log_partPt log_partP log_ptB log_IPs partlcs log_eOverP ghostProb log_IPPU
--- DataSetInfo              :       mult:  +1.000     -0.007    +0.040  +0.069  +0.020  +0.123     +0.119    +0.133   -0.047
--- DataSetInfo              : log_partPt:  -0.007     +1.000    +0.529  +0.073  +0.050  -0.050     -0.068    -0.067   +0.094
--- DataSetInfo              :  log_partP:  +0.040     +0.529    +1.000  +0.052  -0.118  +0.091     -0.032    +0.177   -0.141
--- DataSetInfo              :    log_ptB:  +0.069     +0.073    +0.052  +1.000  -0.005  +0.028     +0.020    +0.033   -0.010
--- DataSetInfo              :    log_IPs:  +0.020     +0.050    -0.118  -0.005  +1.000  -0.027     -0.038    -0.041   -0.049
--- DataSetInfo              :    partlcs:  +0.123     -0.050    +0.091  +0.028  -0.027  +1.000     +0.026    +0.747   -0.093
--- DataSetInfo              : log_eOverP:  +0.119     -0.068    -0.032  +0.020  -0.038  +0.026     +1.000    +0.017   -0.070
--- DataSetInfo              :  ghostProb:  +0.133     -0.067    +0.177  +0.033  -0.041  +0.747     +0.017    +1.000   -0.146
--- DataSetInfo              :   log_IPPU:  -0.047     +0.094    -0.141  -0.010  -0.049  -0.093     -0.070    -0.146   +1.000
--- DataSetInfo              : ----------------------------------------------------------------------------------------------
--- DataSetFactory           :  
--- Factory                  : 
--- Factory                  : current transformation string: 'I,D,N'
--- Factory                  : Create Transformation "I" with events from all classes.
--- Id                       : Transformation, Variable selection : 
--- Id                       : Input : variable 'mult' (index=0).   <---> Output : variable 'mult' (index=0).
--- Id                       : Input : variable 'log_partPt' (index=1).   <---> Output : variable 'log_partPt' (index=1).
--- Id                       : Input : variable 'log_partP' (index=2).   <---> Output : variable 'log_partP' (index=2).
--- Id                       : Input : variable 'log_ptB' (index=3).   <---> Output : variable 'log_ptB' (index=3).
--- Id                       : Input : variable 'log_IPs' (index=4).   <---> Output : variable 'log_IPs' (index=4).
--- Id                       : Input : variable 'partlcs' (index=5).   <---> Output : variable 'partlcs' (index=5).
--- Id                       : Input : variable 'log_eOverP' (index=6).   <---> Output : variable 'log_eOverP' (index=6).
--- Id                       : Input : variable 'ghostProb' (index=7).   <---> Output : variable 'ghostProb' (index=7).
--- Id                       : Input : variable 'log_IPPU' (index=8).   <---> Output : variable 'log_IPPU' (index=8).
--- Factory                  : Create Transformation "D" with events from all classes.
--- Deco                     : Transformation, Variable selection : 
--- Deco                     : Input : variable 'mult' (index=0).   <---> Output : variable 'mult' (index=0).
--- Deco                     : Input : variable 'log_partPt' (index=1).   <---> Output : variable 'log_partPt' (index=1).
--- Deco                     : Input : variable 'log_partP' (index=2).   <---> Output : variable 'log_partP' (index=2).
--- Deco                     : Input : variable 'log_ptB' (index=3).   <---> Output : variable 'log_ptB' (index=3).
--- Deco                     : Input : variable 'log_IPs' (index=4).   <---> Output : variable 'log_IPs' (index=4).
--- Deco                     : Input : variable 'partlcs' (index=5).   <---> Output : variable 'partlcs' (index=5).
--- Deco                     : Input : variable 'log_eOverP' (index=6).   <---> Output : variable 'log_eOverP' (index=6).
--- Deco                     : Input : variable 'ghostProb' (index=7).   <---> Output : variable 'ghostProb' (index=7).
--- Deco                     : Input : variable 'log_IPPU' (index=8).   <---> Output : variable 'log_IPPU' (index=8).
--- Factory                  : Create Transformation "N" with events from all classes.
--- Norm                     : Transformation, Variable selection : 
--- Norm                     : Input : variable 'mult' (index=0).   <---> Output : variable 'mult' (index=0).
--- Norm                     : Input : variable 'log_partPt' (index=1).   <---> Output : variable 'log_partPt' (index=1).
--- Norm                     : Input : variable 'log_partP' (index=2).   <---> Output : variable 'log_partP' (index=2).
--- Norm                     : Input : variable 'log_ptB' (index=3).   <---> Output : variable 'log_ptB' (index=3).
--- Norm                     : Input : variable 'log_IPs' (index=4).   <---> Output : variable 'log_IPs' (index=4).
--- Norm                     : Input : variable 'partlcs' (index=5).   <---> Output : variable 'partlcs' (index=5).
--- Norm                     : Input : variable 'log_eOverP' (index=6).   <---> Output : variable 'log_eOverP' (index=6).
--- Norm                     : Input : variable 'ghostProb' (index=7).   <---> Output : variable 'ghostProb' (index=7).
--- Norm                     : Input : variable 'log_IPPU' (index=8).   <---> Output : variable 'log_IPPU' (index=8).
--- Id                       : Preparing the Identity transformation...
--- Deco                     : Preparing the Decorrelation transformation...
--- Norm                     : Preparing the transformation.
--- TFHandler_Factory        : ---------------------------------------------------------------------
--- TFHandler_Factory        :   Variable          Mean          RMS   [        Min          Max ]
--- TFHandler_Factory        : ---------------------------------------------------------------------
--- TFHandler_Factory        :       mult:    -0.51515     0.23407   [     -1.0000      1.0000 ]
--- TFHandler_Factory        : log_partPt:    -0.25024     0.31504   [     -1.0000      1.0000 ]
--- TFHandler_Factory        :  log_partP:   -0.087827     0.34366   [     -1.0000      1.0000 ]
--- TFHandler_Factory        :    log_ptB:     0.30190     0.20880   [     -1.0000      1.0000 ]
--- TFHandler_Factory        :    log_IPs:    -0.41874     0.31998   [     -1.0000      1.0000 ]
--- TFHandler_Factory        :    partlcs:    -0.32685     0.29081   [     -1.0000      1.0000 ]
--- TFHandler_Factory        : log_eOverP:    -0.44282     0.21791   [     -1.0000      1.0000 ]
--- TFHandler_Factory        :  ghostProb:    -0.43392     0.17059   [     -1.0000      1.0000 ]
--- TFHandler_Factory        :   log_IPPU:    0.045153     0.53188   [     -1.0000      1.0000 ]
--- TFHandler_Factory        : ---------------------------------------------------------------------
--- TFHandler_Factory        : Plot event variables for Id_Deco_Norm
--- TFHandler_Factory        : Create scatter and profile plots in target-file directory: 
--- TFHandler_Factory        : /mnt/mfs/notebook/analyses/tagging_LHCb/MC/tmpaTvBUn/result.root:/InputVariables_Id_Deco_Norm/CorrelationPlots
--- TFHandler_Factory        :  
--- TFHandler_Factory        : Ranking input variables (method unspecific)...
--- Id_Deco_NormTransforma...: Ranking result (top variable is best ranked)
--- Id_Deco_NormTransforma...: -----------------------------------
--- Id_Deco_NormTransforma...: Rank : Variable   : Separation
--- Id_Deco_NormTransforma...: -----------------------------------
--- Id_Deco_NormTransforma...:    1 : log_partPt : 1.621e-02
--- Id_Deco_NormTransforma...:    2 : mult       : 5.223e-03
--- Id_Deco_NormTransforma...:    3 : log_IPs    : 4.948e-03
--- Id_Deco_NormTransforma...:    4 : log_IPPU   : 4.804e-03
--- Id_Deco_NormTransforma...:    5 : log_eOverP : 4.710e-03
--- Id_Deco_NormTransforma...:    6 : log_partP  : 4.247e-03
--- Id_Deco_NormTransforma...:    7 : ghostProb  : 4.118e-03
--- Id_Deco_NormTransforma...:    8 : log_ptB    : 3.144e-03
--- Id_Deco_NormTransforma...:    9 : partlcs    : 2.761e-03
--- Id_Deco_NormTransforma...: -----------------------------------
--- Factory                  :  
--- Factory                  : Train all methods for Classification ...
--- Factory                  : Train method: REP_Estimator for Classification
--- REP_Estimator            : Begin training
--- REP_Estimator            : Training Network
Error in <TDecompLU::InvertLU>: matrix is singular, 2 diag elements < tolerance of 2.2204e-16
--- REP_Estimator            : Finalizing handling of Regulator terms, trainE=0.24448 testE=0.244597
--- REP_Estimator            : Done with handling of Regulator terms
--- REP_Estimator            : End of training                                              
--- REP_Estimator            : Elapsed time for training with 15060 events: 156 sec         
--- REP_Estimator            : Create MVA output for classification on training sample
--- REP_Estimator            : Evaluation of REP_Estimator on training sample (15060 events)
--- REP_Estimator            : Elapsed time for evaluation of 15060 events: 0.118 sec       
--- REP_Estimator            : Creating weight file in xml format: weights/TMVAEstimation_REP_Estimator.weights.xml
--- REP_Estimator            : Creating standalone response class: weights/TMVAEstimation_REP_Estimator.class.C
--- REP_Estimator            : Write special histos to file: /mnt/mfs/notebook/analyses/tagging_LHCb/MC/tmpaTvBUn/result.root:/Method_MLP/REP_Estimator
--- Factory                  : Training finished
--- Factory                  : 
--- Factory                  : Ranking input variables (method specific)...
--- REP_Estimator            : Ranking result (top variable is best ranked)
--- REP_Estimator            : -----------------------------------
--- REP_Estimator            : Rank : Variable   : Importance
--- REP_Estimator            : -----------------------------------
--- REP_Estimator            :    1 : mult       : 1.878e+04
--- REP_Estimator            :    2 : log_IPPU   : 2.941e+02
--- REP_Estimator            :    3 : log_IPs    : 1.217e+02
--- REP_Estimator            :    4 : log_partP  : 9.492e+01
--- REP_Estimator            :    5 : log_ptB    : 5.821e+01
--- REP_Estimator            :    6 : partlcs    : 2.584e+01
--- REP_Estimator            :    7 : log_partPt : 9.020e+00
--- REP_Estimator            :    8 : log_eOverP : 2.041e-01
--- REP_Estimator            :    9 : ghostProb  : 4.910e-02
--- REP_Estimator            : -----------------------------------
--- Factory                  : 
--- Factory                  : === Destroy and recreate all methods via weight files for testing ===
--- Factory                  : 
--- MethodBase               : Reading weight file: weights/TMVAEstimation_REP_Estimator.weights.xml
--- REP_Estimator            : Read method "REP_Estimator" of type "MLP"
--- REP_Estimator            : MVA method was trained with TMVA Version: 4.2.0
--- REP_Estimator            : MVA method was trained with ROOT Version: 5.34/32
--- REP_Estimator            : Building Network
--- REP_Estimator            : Initializing weights

#error "You need a ISO C conforming compiler to use the glibc headers"
*** Interpreter error recovered ***
--- Factory                  : You are running ROOT Version: 5.34/32, Jun 23, 2015
--- Factory                  : 
--- Factory                  : _/_/_/_/_/ _|      _|  _|      _|    _|_|   
--- Factory                  :    _/      _|_|  _|_|  _|      _|  _|    _| 
--- Factory                  :   _/       _|  _|  _|  _|      _|  _|_|_|_| 
--- Factory                  :  _/        _|      _|    _|  _|    _|    _| 
--- Factory                  : _/         _|      _|      _|      _|    _| 
--- Factory                  : 
--- Factory                  : ___________TMVA Version 4.2.0, Sep 19, 2013
--- Factory                  : 
--- DataSetInfo              : Added class "Background"	 with internal class number 0
--- DataSetInfo              : Added class "Signal"	 with internal class number 1
--- Factory                  : Add Tree TrainAssignTree_Background of type Background with 5078 events
--- Factory                  : Add Tree TestAssignTree_Background of type Background with 5078 events
--- Factory                  : Add Tree TrainAssignTree_Signal of type Signal with 9983 events
--- Factory                  : Add Tree TestAssignTree_Signal of type Signal with 9983 events
--- DataSetInfo              : Class index : 0  name : Background
--- DataSetInfo              : Class index : 1  name : Signal
--- Factory                  : Booking method: REP_Estimator
--- REP_Estimator            : Building Network
--- REP_Estimator            : Initializing weights
--- DataSetFactory           : Splitmode is: "RANDOM" the mixmode is: "SAMEASSPLITMODE"
--- DataSetFactory           : Create training and testing trees -- looping over class "Background" ...
--- DataSetFactory           : Weight expression for class 'Background': "weight"
--- DataSetFactory           : Create training and testing trees -- looping over class "Signal" ...
--- DataSetFactory           : Weight expression for class 'Signal': "weight"
--- DataSetFactory           : Number of events in input trees (after possible flattening of arrays):
--- DataSetFactory           :     Background      -- number of events       : 10156  / sum of weights: 10156
--- DataSetFactory           :     Signal          -- number of events       : 19966  / sum of weights: 19966
--- DataSetFactory           :     Background tree -- total number of entries: 10156
--- DataSetFactory           :     Signal     tree -- total number of entries: 19966
--- DataSetFactory           : Preselection: (will NOT affect number of requested training and testing events)
--- DataSetFactory           :     Background requirement: "1"
--- DataSetFactory           :     Background      -- number of events passed: 10156  / sum of weights: 10156
--- DataSetFactory           :     Background      -- efficiency             : 1     
--- DataSetFactory           :     Signal     requirement: "1"
--- DataSetFactory           :     Signal          -- number of events passed: 19966  / sum of weights: 19966
--- DataSetFactory           :     Signal          -- efficiency             : 1     
--- DataSetFactory           : Weight renormalisation mode: "EqualNumEvents": renormalises all event classes ...
--- DataSetFactory           :  such that the effective (weighted) number of events in each class is the same 
--- DataSetFactory           :  (and equals the number of events (entries) given for class=0 )
--- DataSetFactory           : ... i.e. such that Sum[i=1..N_j]{w_i} = N_classA, j=classA, classB, ...
--- DataSetFactory           : ... (note that N_j is the sum of TRAINING events
--- DataSetFactory           :  ..... Testing events are not renormalised nor included in the renormalisation factor!)
--- DataSetFactory           : --> Rescale Background event weights by factor: 1
--- DataSetFactory           : --> Rescale Signal     event weights by factor: 0.508665
--- DataSetFactory           : Number of training and testing events after rescaling:
--- DataSetFactory           : ------------------------------------------------------
--- DataSetFactory           : Background -- training events            : 5078 (sum of weights: 5078) - requested were 0 events
--- DataSetFactory           : Background -- testing events             : 5078 (sum of weights: 5078) - requested were 0 events
--- DataSetFactory           : Background -- training and testing events: 10156 (sum of weights: 10156)
--- DataSetFactory           : Signal     -- training events            : 9983 (sum of weights: 5078) - requested were 0 events
--- DataSetFactory           : Signal     -- testing events             : 9983 (sum of weights: 9983) - requested were 0 events
--- DataSetFactory           : Signal     -- training and testing events: 19966 (sum of weights: 15061)
--- DataSetFactory           : Create internal training tree
--- DataSetFactory           : Create internal testing tree
--- DataSetInfo              : Correlation matrix (Background):
--- DataSetInfo              : ----------------------------------------------------------------------------------------------
--- DataSetInfo              :                mult log_partPt log_partP log_ptB log_IPs partlcs log_eOverP ghostProb log_IPPU
--- DataSetInfo              :       mult:  +1.000     -0.020    +0.027  +0.063  +0.042  +0.101     +0.092    +0.116   -0.071
--- DataSetInfo              : log_partPt:  -0.020     +1.000    +0.515  +0.036  +0.054  -0.069     -0.056    -0.087   +0.068
--- DataSetInfo              :  log_partP:  +0.027     +0.515    +1.000  +0.040  -0.119  +0.086     -0.006    +0.163   -0.163
--- DataSetInfo              :    log_ptB:  +0.063     +0.036    +0.040  +1.000  +0.015  +0.020     +0.022    +0.008   +0.010
--- DataSetInfo              :    log_IPs:  +0.042     +0.054    -0.119  +0.015  +1.000  -0.027     -0.015    -0.012   -0.054
--- DataSetInfo              :    partlcs:  +0.101     -0.069    +0.086  +0.020  -0.027  +1.000     +0.011    +0.731   -0.104
--- DataSetInfo              : log_eOverP:  +0.092     -0.056    -0.006  +0.022  -0.015  +0.011     +1.000    +0.023   -0.077
--- DataSetInfo              :  ghostProb:  +0.116     -0.087    +0.163  +0.008  -0.012  +0.731     +0.023    +1.000   -0.150
--- DataSetInfo              :   log_IPPU:  -0.071     +0.068    -0.163  +0.010  -0.054  -0.104     -0.077    -0.150   +1.000
--- DataSetInfo              : ----------------------------------------------------------------------------------------------
--- DataSetInfo              : Correlation matrix (Signal):
--- DataSetInfo              : ----------------------------------------------------------------------------------------------
--- DataSetInfo              :                mult log_partPt log_partP log_ptB log_IPs partlcs log_eOverP ghostProb log_IPPU
--- DataSetInfo              :       mult:  +1.000     +0.011    +0.019  +0.065  +0.055  +0.068     +0.069    +0.086   -0.031
--- DataSetInfo              : log_partPt:  +0.011     +1.000    +0.568  +0.042  -0.024  -0.054     -0.054    -0.066   +0.075
--- DataSetInfo              :  log_partP:  +0.019     +0.568    +1.000  +0.012  -0.139  +0.058     -0.036    +0.138   -0.118
--- DataSetInfo              :    log_ptB:  +0.065     +0.042    +0.012  +1.000  +0.005  +0.004     +0.005    -0.003   -0.002
--- DataSetInfo              :    log_IPs:  +0.055     -0.024    -0.139  +0.005  +1.000  +0.001     -0.007    -0.015   -0.023
--- DataSetInfo              :    partlcs:  +0.068     -0.054    +0.058  +0.004  +0.001  +1.000     +0.005    +0.728   -0.102
--- DataSetInfo              : log_eOverP:  +0.069     -0.054    -0.036  +0.005  -0.007  +0.005     +1.000    -0.003   -0.066
--- DataSetInfo              :  ghostProb:  +0.086     -0.066    +0.138  -0.003  -0.015  +0.728     -0.003    +1.000   -0.131
--- DataSetInfo              :   log_IPPU:  -0.031     +0.075    -0.118  -0.002  -0.023  -0.102     -0.066    -0.131   +1.000
--- DataSetInfo              : ----------------------------------------------------------------------------------------------
--- DataSetFactory           :  
--- Factory                  : 
--- Factory                  : current transformation string: 'I,D,N'
--- Factory                  : Create Transformation "I" with events from all classes.
--- Id                       : Transformation, Variable selection : 
--- Id                       : Input : variable 'mult' (index=0).   <---> Output : variable 'mult' (index=0).
--- Id                       : Input : variable 'log_partPt' (index=1).   <---> Output : variable 'log_partPt' (index=1).
--- Id                       : Input : variable 'log_partP' (index=2).   <---> Output : variable 'log_partP' (index=2).
--- Id                       : Input : variable 'log_ptB' (index=3).   <---> Output : variable 'log_ptB' (index=3).
--- Id                       : Input : variable 'log_IPs' (index=4).   <---> Output : variable 'log_IPs' (index=4).
--- Id                       : Input : variable 'partlcs' (index=5).   <---> Output : variable 'partlcs' (index=5).
--- Id                       : Input : variable 'log_eOverP' (index=6).   <---> Output : variable 'log_eOverP' (index=6).
--- Id                       : Input : variable 'ghostProb' (index=7).   <---> Output : variable 'ghostProb' (index=7).
--- Id                       : Input : variable 'log_IPPU' (index=8).   <---> Output : variable 'log_IPPU' (index=8).
--- Factory                  : Create Transformation "D" with events from all classes.
--- Deco                     : Transformation, Variable selection : 
--- Deco                     : Input : variable 'mult' (index=0).   <---> Output : variable 'mult' (index=0).
--- Deco                     : Input : variable 'log_partPt' (index=1).   <---> Output : variable 'log_partPt' (index=1).
--- Deco                     : Input : variable 'log_partP' (index=2).   <---> Output : variable 'log_partP' (index=2).
--- Deco                     : Input : variable 'log_ptB' (index=3).   <---> Output : variable 'log_ptB' (index=3).
--- Deco                     : Input : variable 'log_IPs' (index=4).   <---> Output : variable 'log_IPs' (index=4).
--- Deco                     : Input : variable 'partlcs' (index=5).   <---> Output : variable 'partlcs' (index=5).
--- Deco                     : Input : variable 'log_eOverP' (index=6).   <---> Output : variable 'log_eOverP' (index=6).
--- Deco                     : Input : variable 'ghostProb' (index=7).   <---> Output : variable 'ghostProb' (index=7).
--- Deco                     : Input : variable 'log_IPPU' (index=8).   <---> Output : variable 'log_IPPU' (index=8).
--- Factory                  : Create Transformation "N" with events from all classes.
--- Norm                     : Transformation, Variable selection : 
--- Norm                     : Input : variable 'mult' (index=0).   <---> Output : variable 'mult' (index=0).
--- Norm                     : Input : variable 'log_partPt' (index=1).   <---> Output : variable 'log_partPt' (index=1).
--- Norm                     : Input : variable 'log_partP' (index=2).   <---> Output : variable 'log_partP' (index=2).
--- Norm                     : Input : variable 'log_ptB' (index=3).   <---> Output : variable 'log_ptB' (index=3).
--- Norm                     : Input : variable 'log_IPs' (index=4).   <---> Output : variable 'log_IPs' (index=4).
--- Norm                     : Input : variable 'partlcs' (index=5).   <---> Output : variable 'partlcs' (index=5).
--- Norm                     : Input : variable 'log_eOverP' (index=6).   <---> Output : variable 'log_eOverP' (index=6).
--- Norm                     : Input : variable 'ghostProb' (index=7).   <---> Output : variable 'ghostProb' (index=7).
--- Norm                     : Input : variable 'log_IPPU' (index=8).   <---> Output : variable 'log_IPPU' (index=8).
--- Id                       : Preparing the Identity transformation...
--- Deco                     : Preparing the Decorrelation transformation...
--- Norm                     : Preparing the transformation.
--- TFHandler_Factory        : ---------------------------------------------------------------------
--- TFHandler_Factory        :   Variable          Mean          RMS   [        Min          Max ]
--- TFHandler_Factory        : ---------------------------------------------------------------------
--- TFHandler_Factory        :       mult:    -0.51917     0.23191   [     -1.0000      1.0000 ]
--- TFHandler_Factory        : log_partPt:    -0.23983     0.31428   [     -1.0000      1.0000 ]
--- TFHandler_Factory        :  log_partP:    -0.10056     0.35451   [     -1.0000      1.0000 ]
--- TFHandler_Factory        :    log_ptB:     0.29710     0.23392   [     -1.0000      1.0000 ]
--- TFHandler_Factory        :    log_IPs:    -0.39585     0.33462   [     -1.0000      1.0000 ]
--- TFHandler_Factory        :    partlcs:    -0.37139     0.28668   [     -1.0000      1.0000 ]
--- TFHandler_Factory        : log_eOverP:    -0.43432     0.21892   [     -1.0000      1.0000 ]
--- TFHandler_Factory        :  ghostProb:    -0.47080     0.15096   [     -1.0000      1.0000 ]
--- TFHandler_Factory        :   log_IPPU:    0.063401     0.53029   [     -1.0000      1.0000 ]
--- TFHandler_Factory        : ---------------------------------------------------------------------
--- TFHandler_Factory        : Plot event variables for Id_Deco_Norm
--- TFHandler_Factory        : Create scatter and profile plots in target-file directory: 
--- TFHandler_Factory        : /mnt/mfs/notebook/analyses/tagging_LHCb/MC/tmpHXVRWP/result.root:/InputVariables_Id_Deco_Norm/CorrelationPlots
--- TFHandler_Factory        :  
--- TFHandler_Factory        : Ranking input variables (method unspecific)...
--- Id_Deco_NormTransforma...: Ranking result (top variable is best ranked)
--- Id_Deco_NormTransforma...: -----------------------------------
--- Id_Deco_NormTransforma...: Rank : Variable   : Separation
--- Id_Deco_NormTransforma...: -----------------------------------
--- Id_Deco_NormTransforma...:    1 : log_partPt : 1.795e-02
--- Id_Deco_NormTransforma...:    2 : log_eOverP : 7.685e-03
--- Id_Deco_NormTransforma...:    3 : log_IPs    : 6.194e-03
--- Id_Deco_NormTransforma...:    4 : log_IPPU   : 5.758e-03
--- Id_Deco_NormTransforma...:    5 : log_partP  : 4.685e-03
--- Id_Deco_NormTransforma...:    6 : partlcs    : 4.487e-03
--- Id_Deco_NormTransforma...:    7 : mult       : 4.272e-03
--- Id_Deco_NormTransforma...:    8 : ghostProb  : 3.643e-03
--- Id_Deco_NormTransforma...:    9 : log_ptB    : 2.835e-03
--- Id_Deco_NormTransforma...: -----------------------------------
--- Factory                  :  
--- Factory                  : Train all methods for Classification ...
--- Factory                  : Train method: REP_Estimator for Classification
--- REP_Estimator            : Begin training
--- REP_Estimator            : Training Network
Error in <TDecompLU::InvertLU>: matrix is singular, 3 diag elements < tolerance of 2.2204e-16
--- REP_Estimator            : Finalizing handling of Regulator terms, trainE=0.245612 testE=0.245644
--- REP_Estimator            : Done with handling of Regulator terms
--- REP_Estimator            : End of training                                              
--- REP_Estimator            : Elapsed time for training with 15061 events: 161 sec         
--- REP_Estimator            : Create MVA output for classification on training sample
--- REP_Estimator            : Evaluation of REP_Estimator on training sample (15061 events)
--- REP_Estimator            : Elapsed time for evaluation of 15061 events: 0.118 sec       
--- REP_Estimator            : Creating weight file in xml format: weights/TMVAEstimation_REP_Estimator.weights.xml
--- REP_Estimator            : Creating standalone response class: weights/TMVAEstimation_REP_Estimator.class.C
--- REP_Estimator            : Write special histos to file: /mnt/mfs/notebook/analyses/tagging_LHCb/MC/tmpHXVRWP/result.root:/Method_MLP/REP_Estimator
--- Factory                  : Training finished
--- Factory                  : 
--- Factory                  : Ranking input variables (method specific)...
--- REP_Estimator            : Ranking result (top variable is best ranked)
--- REP_Estimator            : -----------------------------------
--- REP_Estimator            : Rank : Variable   : Importance
--- REP_Estimator            : -----------------------------------
--- REP_Estimator            :    1 : mult       : 1.838e+04
--- REP_Estimator            :    2 : log_IPPU   : 1.646e+02
--- REP_Estimator            :    3 : log_partP  : 1.016e+02
--- REP_Estimator            :    4 : log_IPs    : 1.001e+02
--- REP_Estimator            :    5 : log_ptB    : 4.742e+01
--- REP_Estimator            :    6 : partlcs    : 2.508e+01
--- REP_Estimator            :    7 : log_partPt : 9.041e+00
--- REP_Estimator            :    8 : log_eOverP : 1.904e-01
--- REP_Estimator            :    9 : ghostProb  : 4.798e-02
--- REP_Estimator            : -----------------------------------
--- Factory                  : 
--- Factory                  : === Destroy and recreate all methods via weight files for testing ===
--- Factory                  : 
--- MethodBase               : Reading weight file: weights/TMVAEstimation_REP_Estimator.weights.xml
--- REP_Estimator            : Read method "REP_Estimator" of type "MLP"
--- REP_Estimator            : MVA method was trained with TMVA Version: 4.2.0
--- REP_Estimator            : MVA method was trained with ROOT Version: 5.34/32
--- REP_Estimator            : Building Network
--- REP_Estimator            : Initializing weights

In [ ]:

    

# import cPickle
# with open('../models/old-tagging-MC.pkl', 'r') as f:
#     estimators = cPickle.load(f)


    

In [18]:

    

import cPickle
with open('../models/old-tagging-MC.pkl', 'w') as f:
    cPickle.dump(estimators, f)


    

In [16]:

    

estimators.keys()


    

    
Out[16]:





['mu_xgboost',
 'vtx_xgboost',
 'K_xgboost',
 'e_xgboost',
 'mu_tmva',
 'vtx_tmva',
 'K_tmva',
 'e_tmva']

In [17]:

    

for key in datasets.keys():
    for suffix in ['_xgboost', '_tmva']:
        name = key + suffix
        for data, label in zip([datasets[key], datasets_kstar[key], datasets_ks[key]],
                               ['K+-', 'K*', 'Ks']):
            probs = estimators[name].predict_proba(data)[:, 1]        
            print name, label, 'AUC:', roc_auc_score(data['label'].values, 
                                                     probs, sample_weight=data['N_sig_sw'].values)


    

    




KFold prediction using folds column
mu_xgboost K+- AUC: 0.60261930448
KFold prediction using random classifier (length of data passed not equal to length of train)
mu_xgboost K* AUC: 0.611230272746
KFold prediction using random classifier (length of data passed not equal to length of train)
mu_xgboost Ks AUC: 0.607932772058
KFold prediction using folds column
mu_tmva K+- AUC: 0.592940364243
KFold prediction using random classifier (length of data passed not equal to length of train)
mu_tmva K* AUC: 0.598676563222
KFold prediction using random classifier (length of data passed not equal to length of train)
mu_tmva Ks AUC: 0.606136962522
KFold prediction using folds column
vtx_xgboost K+- AUC: 0.588917449491
KFold prediction using random classifier (length of data passed not equal to length of train)
vtx_xgboost K* AUC: 0.587977742307
KFold prediction using random classifier (length of data passed not equal to length of train)
vtx_xgboost Ks AUC: 0.578570099855
KFold prediction using folds column
vtx_tmva K+- AUC: 0.571443978547
KFold prediction using random classifier (length of data passed not equal to length of train)
vtx_tmva K* AUC: 0.568651768274
KFold prediction using random classifier (length of data passed not equal to length of train)
vtx_tmva Ks AUC: 0.567229788736
KFold prediction using folds column
K_xgboost K+- AUC: 0.575600831162
KFold prediction using random classifier (length of data passed not equal to length of train)
K_xgboost K* AUC: 0.574706612609
KFold prediction using random classifier (length of data passed not equal to length of train)
K_xgboost Ks AUC: 0.576090998033
KFold prediction using folds column
K_tmva K+- AUC: 0.550448826315
KFold prediction using random classifier (length of data passed not equal to length of train)
K_tmva K* AUC: 0.548202528866
KFold prediction using random classifier (length of data passed not equal to length of train)
K_tmva Ks AUC: 0.551324697392
KFold prediction using folds column
e_xgboost K+- AUC: 0.590479132787
KFold prediction using random classifier (length of data passed not equal to length of train)
e_xgboost K* AUC: 0.584023324396
KFold prediction using random classifier (length of data passed not equal to length of train)
e_xgboost Ks AUC: 0.590250695631
KFold prediction using folds column
e_tmva K+- AUC: 0.57826002826
KFold prediction using random classifier (length of data passed not equal to length of train)
e_tmva K* AUC: 0.574583626404
KFold prediction using random classifier (length of data passed not equal to length of train)
e_tmva Ks AUC: 0.577786081477

In [18]:

    

from utils import predict_by_estimator, calibrate_probs, estimate_taggers_old_scheme

def combine(estimators, datasets, keys, N_B_events, logistic=False,
            return_calibrations=False, logistic_combined=True, model_name=None, with_roc=False):
    """
    :param suffix: suffix used for taggers
    :param model_name: name for model after combining classifiers
    """
    data_with_predictions = {}
    calibrators_tracks = dict()
    
    # computing calibrated predictions of each tagger
    for key in keys:
        data, probs = predict_by_estimator(estimators[key], [datasets[key]])
        probs_calibrated, calibrators_tracks[key] = \
            calibrate_probs(data.label.values, data.N_sig_sw.values, probs, logistic=logistic)
            
        ids = numpy.array(data['event_id'])
        data_with_predictions[key] = pandas.DataFrame({'prob_{}'.format(key): probs_calibrated, 
                                                       'tag_{}'.format(key): data.tagAnswer.values,
                                                       'weight': data.N_sig_sw.values,
                                                       'signB': data.signB.values}, index=ids)
        
    calibrator_B, table, roc = estimate_taggers_old_scheme(data_with_predictions, keys, N_B_events, 
                                                           model_name=','.join(keys) if model_name is None else model_name, 
                                                           logistic_combined=logistic_combined)
    if return_calibrations and with_roc:
        return table, calibrators_tracks, calibrator_B, roc
    elif return_calibrations and not with_roc:
        return table, calibrators_tracks, calibrator_B
    elif with_roc:
        return table, roc
    else:
        return table


    

In [19]:

    

test_estimators_tmva = dict((key, estimators[key + '_tmva']) for key in ['K', 'mu', 'e', 'vtx'])


    

In [20]:

    

figsize(18, 7)

results_separate_tmva = []
for key in datasets.keys(): 
    x = combine(test_estimators_tmva, datasets, [key], N_B_events, logistic=True)
    results_separate_tmva.append(x)


    

    




KFold prediction using folds column
AUC for tagged: 0.726351149079 AUC with untag: 0.518457825518






    













    













    




Efficiency, not calibrated 0.832908352542
Average AUC 0.726311426806 6.390834128e-05
KFold prediction using folds column
AUC for tagged: 0.671594489405 AUC with untag: 0.532486901488






    













    













    




Efficiency, not calibrated 1.09560516459
Average AUC 0.67142007776 9.7100842861e-05
KFold prediction using folds column
AUC for tagged: 0.67860053712 AUC with untag: 0.551345124444






    













    













    




Efficiency, not calibrated 1.80481556496
Average AUC 0.678432857576 0.000109161995698
KFold prediction using folds column
AUC for tagged: 0.698873776807 AUC with untag: 0.506609931918






    













    













    




Efficiency, not calibrated 0.23642703254
Average AUC 0.698750282162 0.000150610894499

In [21]:

    

pandas.concat(results_separate_tmva)


    

    
Out[21]:






  

    

      

      
name
      
$\epsilon_{tag}, \%$
      
$\Delta \epsilon_{tag}, \%$
      
$D^2$
      
$\Delta D^2$
      
$\epsilon, \%$
      
$\Delta \epsilon, \%$
      
AUC, with untag
      
$\Delta$ AUC, with untag
    
  
  

    

      
0
      
mu
      
5.053829
      
0.018424
      
0.157923
      
0.001368
      
0.798115
      
0.007499
      
51.845783
      
0
    
    

      
0
      
vtx
      
12.265035
      
0.028701
      
0.089629
      
0.000412
      
1.099299
      
0.005673
      
53.248690
      
0
    
    

      
0
      
K
      
17.759661
      
0.034537
      
0.103643
      
0.000381
      
1.840665
      
0.007651
      
55.134512
      
0
    
    

      
0
      
e
      
2.023049
      
0.011657
      
0.122208
      
0.001247
      
0.247233
      
0.002897
      
50.660993
      
0
    
  

In [22]:

    

test_estimators = dict((key, estimators[key + '_xgboost']) for key in ['K', 'mu', 'e', 'vtx'])


    

In [23]:

    

figsize(18, 7)

calibrators_B = dict()
calibrator_tracks = dict()

results_separate = []
for key in datasets.keys(): 
    x, calibrator_tracks[key], calibrators_B[key] = combine(test_estimators, datasets, [key], N_B_events,
                                                            logistic=True, return_calibrations=True)
    results_separate.append(x)


    

    




KFold prediction using folds column
AUC for tagged: 0.730436671744 AUC with untag: 0.518396837178






    













    













    




Efficiency, not calibrated 0.853230911628
Average AUC 0.730399628035 6.71465027744e-05
KFold prediction using folds column
AUC for tagged: 0.679790836959 AUC with untag: 0.532345022729






    













    













    




Efficiency, not calibrated 1.20698414177
Average AUC 0.679777935484 1.67809714232e-05
KFold prediction using folds column
AUC for tagged: 0.6906140029 AUC with untag: 0.551703026024






    













    













    




Efficiency, not calibrated 1.98966266328
Average AUC 0.690598436033 2.14522512864e-05
KFold prediction using folds column
AUC for tagged: 0.704694203363 AUC with untag: 0.50660836551






    













    













    




Efficiency, not calibrated 0.253846873039
Average AUC 0.704601276194 0.000114295549467

In [24]:

    

pandas.concat(results_separate)


    

    
Out[24]:






  

    

      

      
name
      
$\epsilon_{tag}, \%$
      
$\Delta \epsilon_{tag}, \%$
      
$D^2$
      
$\Delta D^2$
      
$\epsilon, \%$
      
$\Delta \epsilon, \%$
      
AUC, with untag
      
$\Delta$ AUC, with untag
    
  
  

    

      
0
      
mu
      
5.053829
      
0.018424
      
0.162459
      
0.001539
      
0.821040
      
0.008334
      
51.839684
      
0
    
    

      
0
      
vtx
      
12.265035
      
0.028701
      
0.098616
      
0.000459
      
1.209523
      
0.006297
      
53.234502
      
0
    
    

      
0
      
K
      
17.759661
      
0.034537
      
0.112161
      
0.000407
      
1.991945
      
0.008195
      
55.170303
      
0
    
    

      
0
      
e
      
2.023049
      
0.011657
      
0.128742
      
0.001369
      
0.260452
      
0.003149
      
50.660837
      
0
    
  

In [25]:

    

from utils import estimate_new_data_old_scheme


    

In [26]:

    

results_separate_kstar = []

for key in ['K', 'e', 'mu', 'vtx']:
    print key
    x = estimate_new_data_old_scheme(test_estimators, datasets_kstar, [key],
                                     calibrator_tracks[key], calibrators_B[key], 
                                     N_B_events_kstar, model_name='K* ' + key)
    results_separate_kstar.append(x)


    

    




K
KFold prediction using random classifier (length of data passed not equal to length of train)






    













    




e
KFold prediction using random classifier (length of data passed not equal to length of train)






    













    




mu
KFold prediction using random classifier (length of data passed not equal to length of train)






    













    




vtx
KFold prediction using random classifier (length of data passed not equal to length of train)






    

In [27]:

    

results_separate_ks = []

for key in ['K', 'e', 'mu', 'vtx']:
    print key
    x = estimate_new_data_old_scheme(test_estimators, datasets_ks, [key],
                                     calibrator_tracks[key], calibrators_B[key], 
                                     N_B_events_ks, model_name='Ks ' + key)
    results_separate_ks.append(x)


    

    




K
KFold prediction using random classifier (length of data passed not equal to length of train)






    













    




e
KFold prediction using random classifier (length of data passed not equal to length of train)






    













    




mu
KFold prediction using random classifier (length of data passed not equal to length of train)






    













    




vtx
KFold prediction using random classifier (length of data passed not equal to length of train)






    

In [28]:

    

pandas.concat(results_separate_kstar + results_separate_ks)


    

    
Out[28]:






  

    

      

      
name
      
$\epsilon_{tag}, \%$
      
$\Delta \epsilon_{tag}, \%$
      
$D^2$
      
$\Delta D^2$
      
$\epsilon, \%$
      
$\Delta \epsilon, \%$
      
AUC, with untag
      
$\Delta$ AUC, with untag
    
  
  

    

      
0
      
K* K
      
17.949948
      
0.064485
      
0.113479
      
0
      
2.036935
      
0.007318
      
55.163053
      
0
    
    

      
0
      
K* e
      
2.065713
      
0.021876
      
0.126489
      
0
      
0.261291
      
0.002767
      
50.646037
      
0
    
    

      
0
      
K* mu
      
5.049726
      
0.034203
      
0.162053
      
0
      
0.818324
      
0.005543
      
51.701056
      
0
    
    

      
0
      
K* vtx
      
12.413041
      
0.053625
      
0.099109
      
0
      
1.230244
      
0.005315
      
53.279870
      
0
    
    

      
0
      
Ks K
      
17.390640
      
0.115273
      
0.115415
      
0
      
2.007135
      
0.013304
      
54.866625
      
0
    
    

      
0
      
Ks e
      
1.969819
      
0.038796
      
0.131601
      
0
      
0.259231
      
0.005106
      
50.636061
      
0
    
    

      
0
      
Ks mu
      
4.935244
      
0.061408
      
0.165429
      
0
      
0.816434
      
0.010159
      
51.820314
      
0
    
    

      
0
      
Ks vtx
      
11.982426
      
0.095685
      
0.102648
      
0
      
1.229967
      
0.009822
      
52.950957
      
0
    
  

In [29]:

    

figsize(18, 7)

x, calibrator_tracks_comb_tmva, calibrator_B_comb_tmva, roc_curve_old_tmva = \
    combine(test_estimators_tmva, datasets,
             ['K', 'e', 'mu', 'vtx'], 
             N_B_events, logistic=True,
             return_calibrations=True, logistic_combined=False,
             model_name='tmva combination', with_roc=True)
results_separate_tmva.append(x)


    

    




KFold prediction using folds column
KFold prediction using folds column
KFold prediction using folds column
KFold prediction using folds column
AUC for tagged: 0.689765360619 AUC with untag: 0.577919971799






    













    













    




Efficiency, not calibrated 4.33568047738
Average AUC 0.68969503732 5.86993113305e-05

In [30]:

    

figsize(18, 7)

x, calibrator_tracks_comb, calibrator_B_comb, roc_curve_old = combine(test_estimators, datasets, 
                                                                      ['K', 'e', 'mu', 'vtx'], N_B_events,
                                                                      logistic=True, return_calibrations=True,
                                                                      logistic_combined=False,
                                                                      model_name='xgboost combination', with_roc=True)
results_separate.append(x)


    

    




KFold prediction using folds column
KFold prediction using folds column
KFold prediction using folds column
KFold prediction using folds column
AUC for tagged: 0.69647990046 AUC with untag: 0.578715078214






    













    













    




Efficiency, not calibrated 4.63029956468
Average AUC 0.696311810402 4.00344798783e-05

In [31]:

    

pandas.concat(results_separate)


    

    
Out[31]:






  

    

      

      
name
      
$\epsilon_{tag}, \%$
      
$\Delta \epsilon_{tag}, \%$
      
$D^2$
      
$\Delta D^2$
      
$\epsilon, \%$
      
$\Delta \epsilon, \%$
      
AUC, with untag
      
$\Delta$ AUC, with untag
    
  
  

    

      
0
      
mu
      
5.053829
      
0.018424
      
0.162459
      
0.001539
      
0.821040
      
0.008334
      
51.839684
      
0
    
    

      
0
      
vtx
      
12.265035
      
0.028701
      
0.098616
      
0.000459
      
1.209523
      
0.006297
      
53.234502
      
0
    
    

      
0
      
K
      
17.759661
      
0.034537
      
0.112161
      
0.000407
      
1.991945
      
0.008195
      
55.170303
      
0
    
    

      
0
      
e
      
2.023049
      
0.011657
      
0.128742
      
0.001369
      
0.260452
      
0.003149
      
50.660837
      
0
    
    

      
0
      
xgboost combination
      
29.043496
      
0.044167
      
0.117465
      
0.000031
      
3.411594
      
0.005267
      
57.871508
      
0
    
  

In [32]:

    

x = estimate_new_data_old_scheme(test_estimators, datasets_kstar, ['K', 'e', 'mu', 'vtx'],
                                 calibrator_tracks_comb, calibrator_B_comb, N_B_events_kstar, 
                                 model_name='K* combination')
results_separate_kstar.append(x)


    

    




KFold prediction using random classifier (length of data passed not equal to length of train)
KFold prediction using random classifier (length of data passed not equal to length of train)
KFold prediction using random classifier (length of data passed not equal to length of train)
KFold prediction using random classifier (length of data passed not equal to length of train)






    

In [33]:

    

x = estimate_new_data_old_scheme(test_estimators_tmva, datasets_kstar, ['K', 'e', 'mu', 'vtx'],
                                 calibrator_tracks_comb_tmva, calibrator_B_comb_tmva, N_B_events_kstar, 
                                 model_name='K* combination')


    

    




KFold prediction using random classifier (length of data passed not equal to length of train)
KFold prediction using random classifier (length of data passed not equal to length of train)
KFold prediction using random classifier (length of data passed not equal to length of train)
KFold prediction using random classifier (length of data passed not equal to length of train)






    

In [34]:

    

x


    

    
Out[34]:






  

    

      

      
name
      
$\epsilon_{tag}, \%$
      
$\Delta \epsilon_{tag}, \%$
      
$D^2$
      
$\Delta D^2$
      
$\epsilon, \%$
      
$\Delta \epsilon, \%$
      
AUC, with untag
      
$\Delta$ AUC, with untag
    
  
  

    

      
0
      
K* combination
      
29.30824
      
0.082399
      
0.111179
      
0
      
3.258474
      
0.009161
      
57.672729
      
0
    
  

In [35]:

    

x = estimate_new_data_old_scheme(test_estimators, datasets_ks, ['K', 'e', 'mu', 'vtx'],
                                 calibrator_tracks_comb, calibrator_B_comb, N_B_events_ks, 
                                 model_name='Ks combination')
results_separate_ks.append(x)


    

    




KFold prediction using random classifier (length of data passed not equal to length of train)
KFold prediction using random classifier (length of data passed not equal to length of train)
KFold prediction using random classifier (length of data passed not equal to length of train)
KFold prediction using random classifier (length of data passed not equal to length of train)






    

In [36]:

    

pandas.concat(results_separate_tmva)


    

    
Out[36]:






  

    

      

      
name
      
$\epsilon_{tag}, \%$
      
$\Delta \epsilon_{tag}, \%$
      
$D^2$
      
$\Delta D^2$
      
$\epsilon, \%$
      
$\Delta \epsilon, \%$
      
AUC, with untag
      
$\Delta$ AUC, with untag
    
  
  

    

      
0
      
mu
      
5.053829
      
0.018424
      
0.157923
      
0.001368
      
0.798115
      
0.007499
      
51.845783
      
0
    
    

      
0
      
vtx
      
12.265035
      
0.028701
      
0.089629
      
0.000412
      
1.099299
      
0.005673
      
53.248690
      
0
    
    

      
0
      
K
      
17.759661
      
0.034537
      
0.103643
      
0.000381
      
1.840665
      
0.007651
      
55.134512
      
0
    
    

      
0
      
e
      
2.023049
      
0.011657
      
0.122208
      
0.001247
      
0.247233
      
0.002897
      
50.660993
      
0
    
    

      
0
      
tmva combination
      
29.043496
      
0.044167
      
0.110931
      
0.000021
      
3.221838
      
0.004937
      
57.791997
      
0
    
  

In [37]:

    

pandas.concat(results_separate)


    

    
Out[37]:






  

    

      

      
name
      
$\epsilon_{tag}, \%$
      
$\Delta \epsilon_{tag}, \%$
      
$D^2$
      
$\Delta D^2$
      
$\epsilon, \%$
      
$\Delta \epsilon, \%$
      
AUC, with untag
      
$\Delta$ AUC, with untag
    
  
  

    

      
0
      
mu
      
5.053829
      
0.018424
      
0.162459
      
0.001539
      
0.821040
      
0.008334
      
51.839684
      
0
    
    

      
0
      
vtx
      
12.265035
      
0.028701
      
0.098616
      
0.000459
      
1.209523
      
0.006297
      
53.234502
      
0
    
    

      
0
      
K
      
17.759661
      
0.034537
      
0.112161
      
0.000407
      
1.991945
      
0.008195
      
55.170303
      
0
    
    

      
0
      
e
      
2.023049
      
0.011657
      
0.128742
      
0.001369
      
0.260452
      
0.003149
      
50.660837
      
0
    
    

      
0
      
xgboost combination
      
29.043496
      
0.044167
      
0.117465
      
0.000031
      
3.411594
      
0.005267
      
57.871508
      
0
    
  

In [38]:

    

pandas.concat(results_separate_kstar)


    

    
Out[38]:






  

    

      

      
name
      
$\epsilon_{tag}, \%$
      
$\Delta \epsilon_{tag}, \%$
      
$D^2$
      
$\Delta D^2$
      
$\epsilon, \%$
      
$\Delta \epsilon, \%$
      
AUC, with untag
      
$\Delta$ AUC, with untag
    
  
  

    

      
0
      
K* K
      
17.949948
      
0.064485
      
0.113479
      
0
      
2.036935
      
0.007318
      
55.163053
      
0
    
    

      
0
      
K* e
      
2.065713
      
0.021876
      
0.126489
      
0
      
0.261291
      
0.002767
      
50.646037
      
0
    
    

      
0
      
K* mu
      
5.049726
      
0.034203
      
0.162053
      
0
      
0.818324
      
0.005543
      
51.701056
      
0
    
    

      
0
      
K* vtx
      
12.413041
      
0.053625
      
0.099109
      
0
      
1.230244
      
0.005315
      
53.279870
      
0
    
    

      
0
      
K* combination
      
29.308240
      
0.082399
      
0.117617
      
0
      
3.447158
      
0.009692
      
57.767091
      
0
    
  

In [39]:

    

pandas.concat(results_separate_ks)


    

    
Out[39]:






  

    

      

      
name
      
$\epsilon_{tag}, \%$
      
$\Delta \epsilon_{tag}, \%$
      
$D^2$
      
$\Delta D^2$
      
$\epsilon, \%$
      
$\Delta \epsilon, \%$
      
AUC, with untag
      
$\Delta$ AUC, with untag
    
  
  

    

      
0
      
Ks K
      
17.390640
      
0.115273
      
0.115415
      
0
      
2.007135
      
0.013304
      
54.866625
      
0
    
    

      
0
      
Ks e
      
1.969819
      
0.038796
      
0.131601
      
0
      
0.259231
      
0.005106
      
50.636061
      
0
    
    

      
0
      
Ks mu
      
4.935244
      
0.061408
      
0.165429
      
0
      
0.816434
      
0.010159
      
51.820314
      
0
    
    

      
0
      
Ks vtx
      
11.982426
      
0.095685
      
0.102648
      
0
      
1.229967
      
0.009822
      
52.950957
      
0
    
    

      
0
      
Ks combination
      
28.387393
      
0.147277
      
0.119903
      
0
      
3.403736
      
0.017659
      
57.402882
      
0
    
  

In [40]:

    

pandas.concat(results_separate_tmva + results_separate + results_separate_kstar + results_separate_ks).to_csv(
    '../img/old-tagging-MC.csv', header=True, index=False)


    

In [41]:

    

import cPickle
with open('../models/old-rocs-MC', 'w') as f:
    cPickle.dump(roc_curve_old, f)


    

In [42]:

    

from utils import prepare_for_epm_old_scheme


    

In [43]:

    

kstar_flavour = pandas.DataFrame(root_numpy.root2array('../datasets/MC/csv/WG/Bd_JPsiKstar/2012/Tracks.root',
                                                       branches=['run', 'event', 'K_MCID']))
kstar_flavour['event_id'] = kstar_flavour.run.apply(str) + '_' + kstar_flavour.event.apply(int).apply(str)
_, ids = numpy.unique(kstar_flavour.event_id, return_index=True)
kstar_flavour = kstar_flavour.loc[ids, :]


    

In [44]:

    

kstar_flavour.index = kstar_flavour.event_id
vals = kstar_flavour.loc[datasets_kstar['vtx'].event_id.values, 'K_MCID']
datasets_kstar['vtx']['K_MCID'] = vals.values


    

In [45]:

    

prepared_kstar = prepare_for_epm_old_scheme(test_estimators, datasets_kstar, ['K', 'e', 'mu', 'vtx'],
                                            calibrator_tracks_comb, calibrator_B_comb, N_B_events_kstar)
root_numpy.array2root(prepared_kstar.to_records(index=False), "../EPM/kstar_MC_old.root", 
                      mode='recreate')


    

    




KFold prediction using random classifier (length of data passed not equal to length of train)
KFold prediction using random classifier (length of data passed not equal to length of train)
KFold prediction using random classifier (length of data passed not equal to length of train)
KFold prediction using random classifier (length of data passed not equal to length of train)
eff tag:  0.293082399164
D2: 0.117617354136
eff: 0.0344715763334

In [46]:

    

from utils import compute_mistag


    

In [47]:

    

# remove last line which correspond to all not passed events
mask = prepared_kstar[:-1]


    

In [48]:

    

figure(figsize=(10, 8))
compute_mistag(mask['probs'].values, mask['signB'].values, mask['weight'].values, mask['signB'].values > -100,
               uniform=False, bins=numpy.linspace(10, 90, 9))
p = (2*(mask['probs'].values - 0.5))**3 / 2 + 0.5
compute_mistag(p, mask['signB'].values, mask['weight'].values, mask['signB'].values > -100,
               uniform=False, bins=numpy.linspace(10, 90, 9))


    

In [49]:

    

prepared_kstar_bad = prepared_kstar.copy()
prepared_kstar_bad['probs'] = list(p) + [0.5]
prepared_kstar_bad['mistag'] = list(numpy.minimum(p, 1-p)) + [0.5]
prepared_kstar_bad['tag'] = list(numpy.where(p > 0.5, 1, -1)) + [0]
root_numpy.array2root(prepared_kstar_bad.to_records(index=False), "../EPM/kstar_MC_old_bad_calibration.root", 
                      mode='recreate')


    

In [50]:

    

time_var = mask.decay_time
proba = mask.probs
weight = mask.weight

time_means = numpy.percentile(time_var, numpy.linspace(0, 100, 30))
time_bins = numpy.searchsorted(time_means, time_var)
d2_vs_time = numpy.bincount(time_bins, weights=(2*(proba - 0.5))**2) / numpy.bincount(time_bins, weights=weight)


    

In [51]:

    

errors = []
for i in range(30):
    probs_in_bin = proba[time_bins == i]
    errors.append(numpy.std((2*(probs_in_bin - 0.5))**2) / numpy.sqrt(len(probs_in_bin)))


    

In [52]:

    

time_means_bins = list(time_means[:-1] + (time_means[1:] - time_means[:-1]) / 2) + [max(time_var)]


    

In [53]:

    

errorbar(time_means_bins, d2_vs_time, yerr=errors, fmt='o')
xlabel('time')
ylabel('D2 in time bin')


    

    
Out[53]:





<matplotlib.text.Text at 0x7fd7e63540d0>






    

In [54]:

    

from sklearn.isotonic import IsotonicRegression
labels = (mask.signB > 0) * 1
iso_calibrator = IsotonicRegression(y_min=0, y_max=1, out_of_bounds='clip')
iso_calibrator.fit(numpy.r_[proba, 1-proba], 
                   numpy.r_[labels, 1 - labels],
                   sample_weight=numpy.r_[weight, weight])
proba_self_calibrated = iso_calibrator.transform(proba)


    

In [55]:

    

figure(figsize=(10, 8))
compute_mistag(proba_self_calibrated, mask['signB'].values, mask['weight'].values, mask['signB'].values > -100,
               uniform=False, bins=numpy.linspace(10, 90, 9))


    

In [56]:

    

sum(time_var > 0), len(time_var)


    

    
Out[56]:





(124549, 126514)

In [57]:

    

hist(time_var[mask.signB * mask.flavour == 1], bins=100, alpha=0.4, range=(0, 1), label='non-osc')
hist(time_var[mask.signB * mask.flavour == -1], bins=100, alpha=0.4, range=(0, 1), label='osc')
legend();


    

In [58]:

    

hist(time_var[mask.signB * mask.flavour == 1], bins=100, alpha=0.4, range=(0, 18), label='non-osc')
hist(time_var[mask.signB * mask.flavour == -1], bins=100, alpha=0.4, range=(0, 18), label='osc')
legend();


    

In [59]:

    

numpy.average((2 * (proba_self_calibrated - 0.5))**2), numpy.average((2 * (proba - 0.5))**2)


    

    
Out[59]:





(0.11357603312646862, 0.11761735413565277)

In [60]:

    

prepared_kstar_self = prepared_kstar.copy()
prepared_kstar_self['probs'] = list(proba_self_calibrated) + [0.5]
prepared_kstar_self['mistag'] = list(numpy.minimum(proba_self_calibrated, 1-proba_self_calibrated)) + [0.5]
prepared_kstar_self['tag'] = list(numpy.where(proba_self_calibrated > 0.5, 1, -1)) + [0]
root_numpy.array2root(prepared_kstar_self.to_records(index=False), "../EPM/kstar_MC_old_self_calibration.root", 
                      mode='recreate')


    

In [61]:

    

from utils import CalibrationProcedure
self_calibrator = CalibrationProcedure(symmetrize=True, random_state=54)
self_calibrator.fit(proba, mask.signB, sample_weight=weight)
proba_self_calibrated_holdout = self_calibrator.predict_proba(proba)


    

In [62]:

    

figure(figsize=(10, 8))
compute_mistag(proba_self_calibrated_holdout, mask['signB'].values, mask['weight'].values, mask['signB'].values > -100,
               uniform=False, bins=numpy.linspace(10, 90, 9))


    

In [63]:

    

prepared_kstar_self = prepared_kstar.copy()
prepared_kstar_self['probs'] = list(proba_self_calibrated_holdout) + [0.5]
prepared_kstar_self['mistag'] = list(numpy.minimum(proba_self_calibrated_holdout, 1-proba_self_calibrated_holdout)) + [0.5]
prepared_kstar_self['tag'] = list(numpy.where(proba_self_calibrated_holdout > 0.5, 1, -1)) + [0]
root_numpy.array2root(prepared_kstar_self.to_records(index=False), "../EPM/kstar_MC_old_self_calibration_holdout.root", 
                      mode='recreate')


    

In [64]:

    

prepared_kstar = prepare_for_epm_old_scheme(test_estimators_tmva, datasets_kstar, ['K', 'e', 'mu', 'vtx'],
                                            calibrator_tracks_comb_tmva, calibrator_B_comb_tmva, N_B_events_kstar)
root_numpy.array2root(prepared_kstar.to_records(index=False), "../EPM/kstar_MC_old_tmva.root", 
                      mode='recreate')


    

    




KFold prediction using random classifier (length of data passed not equal to length of train)
KFold prediction using random classifier (length of data passed not equal to length of train)
KFold prediction using random classifier (length of data passed not equal to length of train)
KFold prediction using random classifier (length of data passed not equal to length of train)
eff tag:  0.293082399164
D2: 0.111179438657
eff: 0.0325847366194