using fit_params from pipeline sklearn for training

Question

I am using XGBClassifier from xgboost library in a Pipeline from sklearn but whenever i want to access one of the **fit_params in the way that the library says to do so https://scikit-learn.org/stable/modules/generated/sklearn.pipeline.Pipeline.html#sklearn.pipeline.Pipeline.fit i got keyerrors

xgb_model = XGBClassifier(eval_metric='logloss', use_label_encoder=False)
pipeline = Pipeline([("preproc", preprocesser), ("classifier", xgb_model)])
pipeline.fit(
    X_train, y_train, train_model__eval_set=[(X_valid_transformed, y_valid)]
)

i got

Keyerror: 'train_model'

Answer 1

From the sklearn.pipeline docs:

...
**fit_paramsdict of string -> object
  Parameters passed to the fit method of each step,
  where each parameter name is prefixed such that
  parameter p for step s has key s__p
...

So, for your code, you need:

                                                      |
                                                      |
                                                      v
                                                   ________
                                                  |        |
pipeline = Pipeline([("preproc", preprocesser), ("classifier", xgb_model)])
pipeline.fit(
    X_train, y_train, classifier__eval_set=[(X_valid_transformed, y_valid)]
)                     |________|
                          ^
                          |
                          |