mlrl.testbed_arff.experiments.input.sources.source_arff module¶

Provides classes that allow reading datasets from ARFF files.

class mlrl.testbed_arff.experiments.input.sources.source_arff.ArffFileSource(directory: Path)¶

Allows to read a dataset from an ARFF file.

class ArffDataset(arff_file: ArffFile, output_names: set[str] | None)¶

Provides access to the content of an ARFF file and the corresponding Mulan XML file, if available.

property feature_matrix: sparray¶: The feature matrix contained in the dataset.

property features: list[Attribute]¶: A list that stores all features contained in the dataset.

static from_file(arff_file: ArffFile, file_path: Path) → ArffDataset¶

Creates and returns an ARFF dataset from given ARFF file and a corresponding Mulan XML file, if available.

Parameters:

Returns:

The ARFF dataset that has been created

property output_matrix: sparray¶: The output matrix contained in the dataset.

property outputs: list[Attribute]¶: A list that stores all outputs contained in the dataset.

property outputs_at_start: bool¶: True, if the outputs are defined before the features, False otherwise.

class ArffFile(matrix: sparray, arff_attributes: list[Any], relation: str)¶

Provides access to the content of an ARFF file.

property attributes: list[Attribute]¶: A list that contains all attributes defined in the ARFF file.

static from_file(file_path: Path, sparse: bool, dtype: dtype) → ArffFile¶

Loads the content of an ARFF file.

Parameters:

file_path – The path to the ARFF file
sparse – True, if the ARFF file is given in sparse format, False otherwise. If the given format is incorrect, an arff.BadLayout is raised
dtype – The type of the data matrix to be read from the file

Returns:

An ArffFileSource.ArffFile that has been loaded

mlrl.testbed_arff.experiments.input.sources.source_arff.normalize_attribute_name(name: str) → str¶

Normalizes the name of an attribute by removing forbidden characters.