module change_detection.evaluation.measures.mean_time_ratio

Mean Time Ratio Measure.

This function returns the mean time ration measure as introduced in: Bifet, Albert, et al. "CD-MOA: Change detection framework for massive online analysis." International Symposium on Intelligent Data Analysis. Springer, Berlin, Heidelberg, 2013.

Copyright (C) 2022 Johannes Haug.

function mean_time_ratio

mean_time_ratio(
    evaluator: float.change_detection.evaluation.change_detection_evaluator.ChangeDetectionEvaluator,
    drifts: list,
    n_delay: int
) → float

Calculates the mean time ratio.

The mean time ratio is a function of the detection delay, the time between false alarms, and the detected change rate: mean_time_ratio = (time_between_false_alarms / detection_delay) * detected_change_rate

Args:

  • evaluator: The ChangeDetectionEvaluator object.
  • drifts: List of time steps corresponding to detected concept drifts.
  • n_delay: The number of observations after a known concept drift, during which we count the detections made by the model as true positives.

Returns:

  • float: The mean time ratio.

This file was automatically generated via lazydocs.