from sklearn.feature_selection import SelectKBest,SelectPercentile


features_list = ['poi','salary', 'deferral_payments', 'total_payments',
       'exercised_stock_options', 'bonus', 'restricted_stock',
       'restricted_stock_deferred', 'total_stock_value', 'expenses',
       'loan_advances', 'director_fees', 'deferred_income',
       'long_term_incentive']

kbest = SelectKBest(k= 4)
kbest.fit(features_train, labels_train)
kbest_financial_feats = zip(features_list[1:], kbest.scores_, kbest.pvalues_, kbest.get_support())
print sorted(kbest_financial_feats, key= lambda kbest_financial_feats: kbest_financial_feats[1], reverse= True)


# output
[('loan_advances', 2.7194355225362137, 0.10230285546818838, True), ('director_fees', 0.44209441196557453, 0.5076592317140044, True), ('total_payments', 0.38222224033137525, 0.53783608133444827, True), ('restricted_stock_deferred', 0.31098646703321225, 0.57833429274585102, True), ('deferral_payments', 0.13733982626793317, 0.71173179458219971, False), ('exercised_stock_options', 0.12830042038876935, 0.72096339272865384, False), ('total_stock_value', 0.095489886234481833, 0.75795996978801217, False), ('deferred_income', 0.054988982219314003, 0.81508322871264716, False), ('bonus', 0.042201153548052976, 0.83765862883975917, False), ('expenses', 0.038486299696967666, 0.84487199943396152, False), ('restricted_stock', 0.019093982688226523, 0.89037805973110706, False), ('salary', 0.007628120832847236, 0.93057838078076993, False), ('long_term_incentive', 0.0058908601173862964, 0.93897578396702075, False)]