from sklearn.datasets import make_classification
from sklearn.grid_search import GridSearchCV
from sklearn.tree import DecisionTreeClassifier
from sklearn.cross_validation import KFold


X, y = make_classification(random_state=0)

est_parameters = {
    "max_depth": range(2, 7)}

def best_est(X, y):
    cv = KFold(y.shape[0], n_folds=2, random_state=0)
    # default tree uses gini and best split
    est = DecisionTreeClassifier()
    grid_search = GridSearchCV(est, est_parameters, cv=cv)
    grid_search.fit(X, y)
    return grid_search.best_score_, grid_search.best_params_

for i in xrange(10):
    print(best_est(X, y))
    # why is the best tree different each time
    # unless I set the random_state in DecisionTreeClassifier() ?