4.2.RandomForestforBNP.py

import sklearn.grid_search as gs

# Fit a random forest on the training set. 
from sklearn import ensemble
rf = ensemble.RandomForestClassifier()
rf.fit(x_train, y_train) # fit 
#print "- The training error is: %.5f" %(1-rf.score(x_train, y_train))
#print "- The test     error is: %.5f" %(1-rf.score(x_test, y_test))

# Find the best parameters using grid_search.GridSearchCV to 
# The initial combination of the parameters: 
# grid_para_forest = {'criterion': ['gini', 'entropy'], 'max_depth': range(1, 31), "n_estimators": range(10, 110, 10)}
grid_para_forest = [{'criterion': ['gini', 'entropy'], 'max_depth': range(1, 31), "n_estimators": range(10, 110, 10)}]
grid_search_forest = gs.GridSearchCV(rf, grid_para_forest, scoring='accuracy', cv=5)
grid_search_forest.fit(x_train, y_train)