Run Random Forest on all data: training and the best test result so far
In [24]:
%matplotlib inline
from matplotlib import pylab as pl
import cPickle as pickle
import pandas as pd
import numpy as np
import os
import random
In [25]:
import sys
sys.path.append('..')
uncommoent the relevant pipeline in ../seizure_detection.py and run
cd ..
./doall data
or
./doall td
./doall tt
In [26]:
FEATURES = 'gen-8_medianwindow-bands2-usf-w60-b0.2-b4-b8-b12-b30-b70-0.1-0.5-0.9'
In [27]:
from common.data import CachedDataLoader
cached_data_loader = CachedDataLoader('../data-cache')
In [28]:
def read_data(target, data_type, features):
fname = 'data_%s_%s_%s'%(data_type,target,features)
print fname
return cached_data_loader.load(fname,None)
In [29]:
best = pd.read_csv('../submissions/141029-predict.10.csv', index_col='clip', squeeze=True)
In [30]:
def prb2logit(x):
return np.log(x/(1.-x))
def logit2prb(x):
return 1./(1+np.exp(-x))
In [31]:
prb2logit(np.clip(best,0.02,0.95)).hist(bins=50)
Out[31]:
In [33]:
from sklearn.ensemble import RandomForestRegressor
from sklearn.cross_validation import StratifiedKFold
from sklearn.metrics import roc_auc_score
clf = RandomForestRegressor(n_estimators=3000, min_samples_split=1, bootstrap=False,max_depth=10,
n_jobs=-1)#, max_features=15
In [34]:
fpout = open('../submissions/141101-predict.4.csv','w')
print >>fpout,'clip,preictal'
In [35]:
def prb2logit(x):
return np.log(x/(1.-x))
def logit2prb(x):
return 1./(1+np.exp(-x))
SMOOTH = 0.
TRAIN_LOGIT = 5.
TEST_MIN = 0.05
TEST_MAX = 0.9
for target in ['Dog_1', 'Dog_2', 'Dog_3', 'Dog_4', 'Dog_5', 'Patient_1', 'Patient_2']:
pdata0 = read_data(target, 'preictal', FEATURES) # positive examples
ndata0 = read_data(target, 'interictal', FEATURES) # negative examples
X0 = np.concatenate((pdata0.X, ndata0.X))
y0 = np.zeros(X0.shape[0])
y0[:pdata0.X.shape[0]] = 1
y0logit = (y0*2.-1.)*TRAIN_LOGIT # turn real test from 1/0 to +/-TRAIN_LOGIT
# predict
tdata = read_data(target, 'test', FEATURES) # test examples
Xt = tdata.X
Nt = Xt.shape[0]
yt = np.array([best['%s_test_segment_%04d.mat' % (target, i+1)] for i in range(Nt)])
yt = prb2logit(np.clip(yt, TEST_MIN, TEST_MAX))
yt = yt*(1.-SMOOTH) + SMOOTH*prb2logit(y0.mean())
X = np.concatenate((X0,Xt))
y = np.concatenate((y0logit, yt))
clf.fit(X,y)
y_proba_logit = clf.predict(Xt)
y_proba = logit2prb(y_proba_logit)
# write results
for i,p in enumerate(y_proba):
print >>fpout,'%s_test_segment_%04d.mat,%.15f' % (target, i+1, p)
In [36]:
fpout.close()
In [37]:
!head ../submissions/141101-predict.4.csv
In [ ]: