In [1]:

    

%matplotlib inline
from matplotlib import pylab as pl
import cPickle as pickle
import pandas as pd
import numpy as np
import os


    

In [10]:

    

df = pd.DataFrame()
df['svm'] = pd.read_csv('../submissions/141116-predict.6.csv', index_col='clip', squeeze=True) #64
df['gb'] = pd.read_csv('../submissions/141107-predict.4.csv', index_col='clip', squeeze=True) #64
df['rf'] = pd.read_csv('../submissions/141106-predict.1.csv', index_col='clip', squeeze=True)
df['rf1'] = pd.read_csv('../submissions/140926-predict.2.csv', index_col='clip', squeeze=True)
df['dbn'] = pd.read_csv('../submissions/140930-predict.5.csv', index_col='clip', squeeze=True)
df['best'] = pd.read_csv('../submissions/141112-predict.1.csv', index_col='clip', squeeze=True)


    

In [4]:

    

pd.scatter_matrix(df[['svm','gb','rf','rf1','dbn','best']],figsize=(6, 6), diagonal='kde');


    

In [18]:

    

w_svm = 1.5
w_gb  = 0.4
w_rf = 0.9
w_rf1 = 0.8
w_dbn = 0.6

s = w_svm + w_gb + w_rf + w_rf1 + w_dbn

w_svm /= s
w_gb /= s
w_rf /= s
w_rf1 /= s
w_dbn /= s


    

In [19]:

    

w_svm, w_gb, w_rf, w_rf1, w_dbn


    

    
Out[19]:





(0.3571428571428572,
 0.09523809523809526,
 0.21428571428571433,
 0.19047619047619052,
 0.14285714285714288)

In [20]:

    

df['preictal'] = w_svm * df['svm'] + w_gb * df['gb'] + w_rf * df['rf'] + w_rf1 * df['rf1'] + w_dbn * df['dbn']


    

In [21]:

    

df['preictal'].to_csv('../submissions/141116-predict.9.csv', header=True)


    

In [22]:

    

pd.scatter_matrix(df[['best','preictal']])


    

    
Out[22]:





array([[<matplotlib.axes._subplots.AxesSubplot object at 0x115788b10>,
        <matplotlib.axes._subplots.AxesSubplot object at 0x11340a310>],
       [<matplotlib.axes._subplots.AxesSubplot object at 0x119298910>,
        <matplotlib.axes._subplots.AxesSubplot object at 0x1132fdf50>]], dtype=object)






    

In [16]: