Open kuraga opened 7 years ago
layer_1 = Ensemble([model_xgboost, model_lightgbm]) layer_2 = Ensemble([XGBClassifier()]) stack = EnsembleStack(cv=4) stack.add_layer(layer_1) stack.add_layer(layer_2) model_stack = EnsembleStackClassifier(stack) y_test_pred_stack = model_stack.predict_proba(X_test)
--------------------------------------------------------------------------- IndexError Traceback (most recent call last) <ipython-input-37-a466b4b341d5> in <module>() ----> 1 y_test_pred_stack = model_stack.predict(X_test) 2 score_prediction(y_test, y_test_pred_stack) /home/sasha/.local/lib64/python3.4/site-packages/brew/stacking/stacker.py in predict(self, X) 85 86 def predict(self, X): ---> 87 out = self.stack.output(X) 88 return self.combiner.combine(out) 89 /home/sasha/.local/lib64/python3.4/site-packages/brew/stacking/stacker.py in output(self, X) 56 57 for layer in self.layers: ---> 58 out = layer.output(input_, mode=self.mode) 59 input_ = out[:, 1:, :].reshape( 60 out.shape[0], (out.shape[1] - 1) * out.shape[2]) /home/sasha/.local/lib64/python3.4/site-packages/brew/base.py in output(self, X, mode) 187 if mode == 'probs': 188 probas = np.zeros((X.shape[0], n_classes)) --> 189 probas[:, list(c.classes_)] = c.predict_proba(X) 190 out[:, :, i] = probas 191 IndexError: only integers, slices (`:`), ellipsis (`...`), numpy.newaxis (`None`) and integer or boolean arrays are valid indices
Python 3.4. list(model_xgboost.classes_) gives [0.0, 1.0]. Why not just probas =?
list(model_xgboost.classes_)
[0.0, 1.0]
probas =
Python 3.4.
list(model_xgboost.classes_)
gives[0.0, 1.0]
. Why not justprobas =
?