Core

Config

add_defaults(config: dict, default: str | None = None) → dict

Add default config values into a given config dict.

Parameters:

• config (dict) – The base config. Modified in place.

• default (str) – The config with default values. None means DEFAULT_CONFIG.

Returns:

The updated config dict.

Return type:

dict

Paths

class CaTabRaPaths

Bases: object

Data class defining file names of objects generated by CaTabRa

Config = 'config.json'

Configuration file as in core.paths.

TrainData = 'train_data.h5'

Data used for training.

TestData = 'test_data.h5'

Data used for testing.

ExplanationData = 'explanation_data.h5'

Data on which predictions were explained.

Statistics = 'statistics'

Folder for descriptive statistics of data set.

Encoder = 'encoder.json'

Encoder parameters that can be loaded as Encoder object.

ModelSummary = 'model_summary.json'

Summary of models created during training.

Model = 'model.joblib'

The final (best) model from the training pipeline.

TrainingHistory = 'training_history.xlsx'

History of the training as performance over time.

OODModel = 'ood.joblib'

Out-of-distribution detector generated during training.

OODStats = 'ood.xlsx'

Probabilities and decisions of data being OOD,

ConsoleLogs = 'console.txt'

Console output redirected to log file.

Invocation = 'invocation.json'

Invocation file containing the parameters used in call.

Predictions = 'predictions.xlsx'

Model decisions for a data set.

FittedEnsemble = 'fitted_ensemble.joblib'

Ensemble fitted during training.

Versions = 'versions.txt'

Versions of Python packages used on creation.

Base

class Invocation(*table: str | Path | DataFrame, split: str | None = None, sample_weight: str | None = None, out: str | Path | None = None, jobs: int | None = None)

Bases: ABC

class CaTabRaBase(invocation: str | Path | dict | None = None)

Bases: ABC