sklearn_model_selection.py 文件源码

def choose_classifier(classifier,  # which classifier to use
                      # parameters for the tree based classifiers
                      trees_n_estimators=None, trees_criterion=None,
                      trees_max_features=None, trees_max_depth=None,
                      # the ones for k-nearest-neighbors
                      knn_n_neighbors=None, knn_weights=None):
    # note that possibly inactive variables have to be optional
    # as ac_pysmac does not assign a value for inactive variables
    # during the minimization phase
    if classifier == 'random_forest':
        predictor = sklearn.ensemble.RandomForestClassifier(
            trees_n_estimators, trees_criterion,
            trees_max_features, trees_max_depth)
    elif classifier == 'extra_trees':
        predictor = sklearn.ensemble.ExtraTreesClassifier(
            trees_n_estimators, trees_criterion,
            trees_max_features, trees_max_depth)
    elif classifier == 'k_nearest_neighbors':
        predictor = sklearn.neighbors.KNeighborsClassifier(
            knn_n_neighbors, knn_weights)

    predictor.fit(X_train, Y_train)
    return -predictor.score(X_test, Y_test)


# defining all the parameters with respective defaults.