Enum API Reference#

All enumeration types are defined in chronocratic.datasets.enums and re-exported from the package root. These enums control loading modes, scaling strategies, and dataset variants across modules and datasets.

Typed enumerations for dataset parameters.

class chronocratic.datasets.enums.ClassificationLoaderMode(*values)

Bases: StrEnum

Loader-level mode for classification datasets.

SAMPLE_ONLY: Returns only the input data without labels.

SAMPLE_LABEL: Returns the input data and its corresponding label.

class chronocratic.datasets.enums.ClassificationSplitMode(*values)

Bases: StrEnum

Strategy for classification train/test data splitting.

AS_DEFINED: Uses the train/test split as defined in the original dataset.

MANUAL: Allows manual specification of train/test split ratios.

class chronocratic.datasets.enums.DataForm(*values)

Bases: StrEnum

Enum for the form (shape) of the data.

REGULAR: 2-D tabular data (samples x features).

NESTED: 3-D array data (samples x timesteps x features).

MULTI_FILES: List of 1-D arrays from multiple files.

class chronocratic.datasets.enums.DataPartition(*values)

Bases: StrEnum

Data partition for train/validation/test splits.

TRAIN: Training data partition.

VAL: Validation data partition.

TEST: Test data partition.

class chronocratic.datasets.enums.ForecastingLoaderMode(*values)

Bases: StrEnum

Loader-level mode for forecasting datasets.

RAW_SERIES: Returns the full time series without splitting.

INPUT_TARGET: Returns paired input and target tensors for supervised forecasting.

INPUT_ONLY: Returns only the input portion of the series.

class chronocratic.datasets.enums.ForecastingMode(*values)

Bases: StrEnum

Whether forecasting is univariate or multivariate.

UNIVARIATE: Uses a single target variable per sample.

MULTIVARIATE: Uses all available variables per sample.

class chronocratic.datasets.enums.ScalingMethod(*values)

Bases: StrEnum

Method for data scaling.

NONE: No scaling applied.

MINMAX: Scales data to a specified range (default 0-1).

STANDARD: Standardizes data to zero mean and unit variance.

class chronocratic.datasets.enums.TimeSeriesDatasetMode(*values)

Bases: StrEnum

Mode for how the dataset yields samples.

SAMPLE_ONLY: Returns only the input data without labels.

SAMPLE_LABEL: Returns the input data and its corresponding label.

INPUT_OUTPUT: Returns separate input and output tensors for supervised learning.