无法运行StackNet分类器的predict_proba

Question

Unable to run predict_proba for Stacknet Classifier.

I constructed StackNet Classifier as below:

Level 0 : XGBClassifier, GradientBoostingRegressor,CatBoostClassifier
Level 1 : XGBClassifier

Model fit is successful. But, I tried to run …..model.predict_proba(Xtrain_prep).

faced below error exception

ValueError: feature_names mismatch

I don't think, its issue with dataset. it worked well with individual classifiers. Appreciate your help on StackNet Classifier.

# Specify model tree for StackNet
models = [[xgb_clf, gbrt_clf, cat_clf], # Level 0
          [xgb_clf]] # Level 1

# Specify parameters for stacked model and begin training
model = StackNetClassifier(models, 
                           metric="auc", 
                           folds=2,
                           restacking=False,
                           use_retraining=True,
                           use_proba=True, # To use predict_proba after training
                           random_state=seed,
                           n_jobs=-1, 
                           verbose=1)
train_preds = model.predict_proba(X_train_prep)[:, 1]

~\Anaconda3\lib\concurrent\futures\_base.py in result(self, timeout)
    430                 raise CancelledError()
    431             elif self._state == FINISHED:
--> 432                 return self.__get_result()
    433             else:
    434                 raise TimeoutError()

~\Anaconda3\lib\concurrent\futures\_base.py in __get_result(self)
    382     def __get_result(self):
    383         if self._exception:
--> 384             raise self._exception
    385         else:
    386             return self._result

ValueError: feature_names mismatch:

Answer 1

found my answer in following github link. https://github.com/dmlc/xgboost/issues/2334

Real Issue

inside fit function, Xtrain is converted to numpy.ndarray, then it is passed to "predictproba" method. so , when we input model.predict_ proba (X_train) …..this input is still in dataframe. Hence, the mismatch while running "predict_proba".

Solution

StackNet wrapper of "predict_proba" should convert the input dataset into numpy.ndarray. or

we can convert to dataset into array as below before passing to "predict_proba": Xmatrix = Xtrain.asmatrix() trainpreds = model.predictproba(Xmatrix)[:, 1]