cheat.py

import numpy as np
import pandas as pd
import csv
from sklearn.tree import DecisionTreeClassifier
from sklearn.ensemble import ExtraTreesClassifier
from sklearn import ensemble
 
print('Load data...')
train = pd.read_csv("./data/train.csv")
target = train['target'].values
train = train.drop(['ID','target','v8','v23','v25','v31','v36','v37','v46','v51','v53','v54','v63','v73','v75','v79','v81','v82','v89','v92','v95','v105','v107','v108','v109','v110','v116','v117','v118','v119','v123','v124','v128'],axis=1)
test = pd.read_csv("./data/test.csv")
id_test = test['ID'].values
test = test.drop(['ID','v8','v23','v25','v31','v36','v37','v46','v51','v53','v54','v63','v73','v75','v79','v81','v82','v89','v92','v95','v105','v107','v108','v109','v110','v116','v117','v118','v119','v123','v124','v128'],axis=1)
#train.loc[:, 'user_type'] = (train.isnull().sum(axis=1) > 60)
#test.loc[:, 'user_type'] = (test.isnull().sum(axis=1) > 60)
print('Clearing...')
for (train_name, train_series), (test_name, test_series) in zip(train.iteritems(),test.iteritems()):
    if train_series.dtype == 'O':
        #for objects: factorize
        train[train_name], tmp_indexer = pd.factorize(train[train_name])
        test[test_name] = tmp_indexer.get_indexer(test[test_name])
        #but now we have -1 values (NaN)
    else:
        #for int or float: fill NaN
        tmp_len = len(train[train_series.isnull()])
        if tmp_len>0:
            #print "mean", train_series.mean()
            train.loc[train_series.isnull(), train_name] = -999
        #and Test
        tmp_len = len(test[test_series.isnull()])
        if tmp_len>0:
            test.loc[test_series.isnull(), test_name] = -999
           
 
 
X_train = train
X_test = test
print('Training...')
extc = ExtraTreesClassifier(n_estimators=2850,max_features= 60,criterion= 'entropy',min_samples_split= 4,
                            max_depth= 40, min_samples_leaf= 2, n_jobs = -1)      
 
extc.fit(X_train,target)
 
print('Predict...')
y_pred = extc.predict_proba(X_test)
#print y_pred
 
pd.DataFrame({"ID": id_test, "PredictedProb": y_pred[:,1]}).to_csv('extra_trees.csv',index=False)