CART_Trainer.py 文件源码

def tree_based_selection(self, data_set, data_target, feature_names):
        """

        :param data_set:
        :return:
        """

        clf = ExtraTreesClassifier()
        clf = clf.fit(data_set, data_target)
        print clf.feature_importances_

        model = SelectFromModel(clf, prefit=True)
        feature_set = model.transform(data_set)

        fea_index = []
        for A_col in np.arange(data_set.shape[1]):
            for B_col in np.arange(feature_set.shape[1]):
                if (data_set[:, A_col] == feature_set[:, B_col]).all():
                    fea_index.append(A_col)

        check = {}
        for i in fea_index:
            check[feature_names[i]] = data_set[0][i]
        print np.array(check)

        return feature_set, fea_index