参数化网格搜索中特征选择步骤的估计器？答案

【问题标题】：Parametrize the estimator of a feature selection step in a grid search?参数化网格搜索中特征选择步骤的估计器？
【发布时间】：2021-12-31 10:02:54
【问题描述】：

如何在网格搜索中参数化特征选择步骤的估计器（此处为 n_estimators 或 RandomForestClassifier）？

from sklearn.feature_selection import SelectFromModel
from sklearn.ensemble import RandomForestClassifier

sfm = SelectFromModel(RandomForestClassifier(n_estimators=100, random_state=42))

pipe = Pipeline(steps=[('preprocessor', preprocessor),
                       ('selector',  sfm),
                       ('regressor', lr)])

from sklearn.model_selection import GridSearchCV

param_grid = {
    "selector__threshold": ['0.45*median',
                            '0.5*median',
                            '0.55*median',
                            '1*median',
                            '1.25*median',
                            '1.5*median',
                            '1.75*median',
                            '2*median'],
    "regressor__penalty": ['l2'],
    "regressor__C": [0.05, 0.1, 0.25, 0.5, 0.75, 1, 1.5],   
}

search = GridSearchCV(pipe,
                      param_grid,
                      n_jobs=-1,
                      refit=True,
                      verbose=3)

search.fit(X_train, y_train)

【问题讨论】：

标签： python scikit-learn feature-selection

【解决方案1】：

做到了：

'selector__estimator__n_estimators': [10, 50, 100],

【讨论】：