y array-like of shape (n_samples,), default=None Parameters X array-like of shape (n_samples, n_features) Get a mask, or integer index, of the features selected.įit the SelectFromModel meta-transformer only once.įit ( X, y = None, ** fit_params ) ¶įit the SelectFromModel meta-transformer. Mask feature names according to selected features. transform ( X ) array(,, , ])įit the SelectFromModel meta-transformer. ] > y = > selector = SelectFromModel ( estimator = LogisticRegression ()). > from sklearn.feature_selection import SelectFromModel > from sklearn.linear_model import LogisticRegression > X =. To only select based on max_features, set threshold=-np.inf. The maximum number of features to select.
Threshold in the case where the coef_ attribute of theĮstimator is of dimension 2. Order of the norm used to filter the vectors of coefficients below norm_order non-zero int, inf, -inf, default=1 Otherwise train the model using fit and then transform to doįeature selection. GridSearchCV and similar utilities that clone the estimator. If True, transform must be called directlyĪnd SelectFromModel cannot be used with cross_val_score, Whether a prefit model is expected to be passed into the constructorĭirectly or not.
Or implicitly (e.g, Lasso), the threshold used is 1e-5. If None and if theĮstimator has a parameter penalty set to l1, either explicitly A scalingįactor (e.g., “1.25*mean”) may also be used. Importance is greater or equal are kept while the others areĭiscarded. The threshold value to use for feature selection. Otherwise, the importance_getter parameter should be used. The estimator should have aįeature_importances_ or coef_ attribute after fitting. This can be both a fitted (if prefit is set to True) The base estimator from which the transformer is built.
0 kommentar(er)
