[FIXED] How do use a GridSearch to search the parameters of a model in a pipeline?

Issue

I need to use grid search to search the parameters of model in my_pipeline How can I do this?

I don't want to search te parameters of the pipeline instead I want to search the parameters of the model that is inside the pipeline

Or_pipeline =  make_pipeline(OR_preprocessor,model)
XGB_model = XGBRegressor()
params = {'learning_rate': [0.1,0.2,0.3,0.4,0.5,0.6,0.7,0.8,0.9,1],
         'n_estimators': [200,400.600,800,1000]}

grid_search_XGB = GridSearchCV(estimator = OR_pipeline ,param_grid= params,scoring = 'neg_mean_absolute_error',
                              n_jobs =1,
                              cv=3)

grid_search_XGB.fit(x_train,y_train)```

 

ValueError: Invalid parameter learning_rate for estimator Pipeline(steps=[('columntransformer',
                 ColumnTransformer(transformers=[('or', OrdinalEncoder(),
                                                  ['country', 'store',
                                                   'product'])])),
                ('xgbregressor',
                 XGBRegressor(base_score=0.5, booster='gbtree', ...))]).

Solution

The usual practice is running grid search on the pipeline, i.e grid_search_XGB = GridSearchCV(estimator = my_pipeline, ...). (See the similar question.) You'll need to supply prefixes to your params names according to pipeline stage naming. This also allows tuning the preprocessing stage parameters if needed.

Answered By - dx2-66