automl.metalearning.metafeatures package

Submodules

automl.metalearning.metafeatures.computation module

Module that exposes the classes to compute statistics on a dataset.

class automl.metalearning.metafeatures.computation.StatisticalInformation

Bases: object

Compute statistics on a dataset.

static class_entropy(y)

Compute statistic.

static class_ocurrences(y)

Compute statistic.

static class_probability_max(y)

Compute statistic.

static class_probability_mean(y)

Compute statistic.

static class_probability_min(y)

Compute statistic.

static class_probability_std(y)

Compute statistic.

static dataset_ratio(X)

Compute statistic.

static inverse_dataset_ratio(X)

Compute statistic.

static kurtosis_max(X, categorical_indicators)

Compute statistic.

static kurtosis_mean(X, categorical_indicators)

Compute statistic.

static kurtosis_min(X, categorical_indicators)

Compute statistic.

static kurtosis_std(X, categorical_indicators)

Compute statistic.

static kurtosisses(X, categorical_indicators)

Compute statistic.

static landmark_1NN(X, y)

Compute statistic.

static landmark_decision_node_learner(X, y)

Compute statistic.

static landmark_decision_tree(X, y)

Compute statistic.

static landmark_lda(X, y)

Compute statistic.

static landmark_naive_bayes(X, y)

Compute statistic.

static landmark_random_node_learner(X, y)

Compute statistic.

static log_dataset_ratio(X)

Compute statistic.

static log_inverse_dataset_ratio(X)

Compute statistic.

static log_number_of_features(X)

Compute statistic.

static log_number_of_instances(X)

Compute statistic.

static missing_values(X)

Compute statistic.

static number_of_categorical_features(categorical_indicators)

Compute statistic.

static number_of_classes(y)

Compute statistic.

static number_of_features(X)

Compute statistic.

static number_of_features_with_na(X)

Compute statistic.

static number_of_instances(X)

Compute statistic.

static number_of_instances_with_na(X)

Compute statistic.

static number_of_missing_values(X)

Compute statistic.

static number_of_numeric_features(categorical_indicators)

Compute statistic.

static number_of_symbols(X, categorical_indicators)

Compute statistic.

static pca(X)

Compute statistic.

static pca_fraction_components_95v(X, pca=None)

Compute statistic.

static pca_kurtosis_first_pc(X, pca=None)

Compute statistic.

static pca_skewness_first_pc(X, pca=None)

Compute statistic.

static percentage_of_features_with_na(X)

Compute statistic.

static percentage_of_instances_with_na(X)

Compute statistic.

static percentage_of_missing_values(X)

Compute statistic.

static ratio_nominal_numerical(categorical_indicators)

Compute statistic.

static ratio_numerical_nominal(categorical_indicators)

Compute statistic.

static skewness_max(X, categorical_indicators)

Compute statistic.

static skewness_mean(X, categorical_indicators)

Compute statistic.

static skewness_min(X, categorical_indicators)

Compute statistic.

static skewness_std(X, categorical_indicators)

Compute statistic.

static skewnesses(X, categorical_indicators)

Compute statistic.

static symbols_max(X, categorical_indicators)

Compute statistic.

static symbols_mean(X, categorical_indicators)

Compute statistic.

static symbols_min(X, categorical_indicators)

Compute statistic.

static symbols_std(X, categorical_indicators)

Compute statistic.

static symbols_sum(X, categorical_indicators)

Compute statistic.

automl.metalearning.metafeatures.metafeatures_interaction module

Module intented to expose the classes to provide dataset metafeatures.

This module includes the global variables to refer the meta-features.

class automl.metalearning.metafeatures.metafeatures_interaction.MetaFeaturesManager(dataset=None)

Bases: object

Class to obtain/interact with the metafeatures for a given dataset.

Parameters:dataset (Dataset) – The dataset to work with. Defaults to None.

dataset

The dataset to work with.

Type:Dataset

metafeatures_as_dict(recompute=False)

Get the dataset’s metafeatures in the form of a dictionary.

Parameters:recompute (bool) – Whether or not to force the recomputing of the metafeatures even if they were already computed. Defaults to False.

metafeatures_as_numpy_array(recompute=False)

Get the dataset’s metafeatures in the form of a numpy darray.

Parameters:recompute (bool) – Whether or not to force the recomputing of the metafeatures even if they were already computed. Defaults to False.

metafeatures_as_pandas_df(recompute=False)

Get the dataset’s metafeatures in the form of a pandas data frame.

Parameters:recompute (bool) – Whether or not to force the recomputing of the metafeatures even if they were already computed. Defaults to False.

Module contents

Compute metafeatures and interact with the precomputed ones.

The methods to compute each field of the metafeatures vector are provided here. Also, the class to interact with the metafeatures is provided.