为什么在将 RandomSearchCV 或 GridSearchCV 与逻辑回归一起使用时出现错误？

Question

pipe=Pipeline(steps=[('standardscaler', StandardScaler()),
                 ('logisticregression', LogisticRegression())])
pipe.fit(X_train,y_train)
random_estimator = RandomizedSearchCV(estimator = pipe,
                               param_distributions = random_grid,
                               n_iter = 100,
                               scoring = 'accuracy',
                               n_jobs = -1,
                               verbose = 1, 
                               random_state = 42,
                              )


random_estimator.fit(X,y)````

i know i am making some noob error but after this i am getting an error

ValueError: Invalid parameter warm_start for estimator Pipeline(steps=[('standardscaler', StandardScaler()), ('logisticregression', LogisticRegression())]). Check the list of available parameters with estimator.get_params().keys() .

after running pipe.get_params().keys() i am getting this

pipe.get_params().keys()

dict_keys(['memory', 'steps', 'verbose', 'standardscaler', 'logisticregression', 'standardscaler__copy', 'standardscaler__with_mean', 'standardscaler__with_std', 'logisticregression__C', 'logisticregression__class_weight', 'logisticregression__dual', 'logisticregression__fit_intercept', 'logisticregression__intercept_scaling', 'logisticregression__l1_ratio', 'logisticregression__max_iter', 'logisticregression__multi_class', 'logisticregression__n_jobs', 'logisticregression__penalty', 'logisticregression__random_state', 'logisticregression__solver', 'logisticregression__tol', 'logisticregression__verbose', 'logisticregression__warm_start'])

can anyone explain the error thank you!!

Answer 1

Your random_grid parameters dictionary currently likely includes the warm_start parameter, something like:

random_grid = {'warm_start': [True, False]}

That name needs to match the logisticregression__warm_start name you see among the pipeline parameter keys in your question:

random_grid = {'logisticregression__warm_start': [True, False]}