model_base.py 文件源码

def greedy_select_features(self):
        saved = None if self.debug_ else self.load('chosen_features')
        if saved == None:
            print('initial shapes:', self.train_.shape, self.test_.shape)
            num_columns = self.train_.shape[1]
            col_names = [str(c) for c in range(num_columns)]
            self.train_.columns = col_names
            self.test_.columns = col_names

            g_best_score = 1e9
            g_best_features = None

            y = self.y_.ravel()
            current = set()
            scorer = metrics.make_scorer(metrics.log_loss)
            for _ in enumerate(col_names):
                avail = set(col_names).difference(current)
                best_score = 1e9
                best_features = None
                for f in avail:
                    newf = list(current | {f})
                    cv = model_selection.cross_val_score(linear_model.BayesianRidge(),
                                                         self.train_[newf], y,
                                                         cv=self.n_fold_, n_jobs=-2,
                                                         scoring = scorer)
                    score = np.mean(cv)
                    if best_score > score:
                        best_score = score
                        best_features = newf
                current = set(best_features)
                if g_best_score > best_score:
                    g_best_score = best_score
                    g_best_features = best_features
                    print('new best:', g_best_score, g_best_features, self.now())
                if len(best_features) - len(g_best_features) > 15:
                    break
            self.save('chosen_features', (g_best_features, None))
        else:
            g_best_features, _ = saved

        print('feature selection complete.', self.now())
        self.train_ = self.train_[g_best_features]
        self.test_ = self.test_[g_best_features]