module feature_selection.evaluation.feature_selection_evaluator

Online Feature Selection Evaluator.

This module contains an evaluator class for online feature selection methods.

Copyright (C) 2022 Johannes Haug.

class FeatureSelectionEvaluator

Online feature selection evaluator class.

Attributes:

  • measure_funcs (List[Callable]): List of evaluation measure functions.
  • decay_rate (float |None): If this parameter is not None, the measurements are additionally aggregated with the specific decay/fading factor.
  • window_size (int | None): If this parameter is not None, the measurements are additionally aggregated in a sliding window.
  • comp_times (list): List of computation times per iteration of feature weighting and selection.
  • memory_changes (list): Memory changes (in GB RAM) per training iteration of the online feature selection model.
  • result (dict): The raw and aggregated measurements of each evaluation measure function.

method FeatureSelectionEvaluator.__init__

__init__(
    measure_funcs: List[Callable],
    decay_rate: Optional[float] = None,
    window_size: Optional[int] = None
)

Inits the online feature selection evaluation object.

Args:

  • measure_funcs: List of evaluation measure functions.
  • decay_rate: If this parameter is not None, the measurements are additionally aggregated with the specific decay/fading factor.
  • window_size: If this parameter is not None, the measurements are additionally aggregated in a sliding window.

method FeatureSelectionEvaluator.run

run(selected_features_history: List[list], n_total_features: int)

Updates relevant statistics and computes the evaluation measures.

Args:

  • selected_features_history: A list of all selected feature vectors obtained over time.
  • n_total_features: The total number of features.

Raises:

  • TypeError: If the calculation of a measure runs an error.

This file was automatically generated via lazydocs.