SequentialFeatureSelection¶
-
class
museotoolbox.ai.SequentialFeatureSelection(classifier, param_grid, path_to_save_models=False, n_comp=1, verbose=False)[source]¶ Sequential Feature Selection
- Parameters
classifier (class.) – Classifier from scikit-learn.
param_grid (np.ndarray.) – param_grid for hyperparameters of the classifier.
path_to_save_models (False or str, optional (default=False)) – If False, will store best model per combination in memory. If str, must be path to save each model and accuracy per feature.
n_comp (int, optional (default=1)) – The number of component per feature. If 4, each feature has 4 columns.
verbose (bool or int, optional (default=False)) – The higher it is the more sequential will show progression.
Methods
__init__(classifier, param_grid[, …])Initialize self.
customize_array(xFunction, **kwargs)fit(X, y[, group, cv, scoring, standardize, …])- param X
shape of np.ndarray is (n_size,n_bands).
get_best_model([clone])predict(X, idx)Predict in raster using the best features.
predict_best_combination(in_image, out_image)Predict in raster using the best features.
predict_images(in_image, out_image_prefix[, …])Predict each best found features with SFFS.fit(X,y).
transform(X[, idx, customizeX])- param idx
The idx to return X array
