algorithms (pmrf.algorithms)

General helper algorithms for anomaly and convergence detection.

Functions

get_anomaly_mask(Y, threshold)	Computes a boolean mask of rows that ARE anomalies.
has_sudden_changes(values, *[, upwards, ...])	Checks if an array has any sudden changes for all values relative to the previous window values.
has_converged_by_absolute_tolerance(values, atol)	Returns True if the last window values are all below the absolute tolerance.
has_converged_by_relative_tolerance(values)	Returns True if the relative change between consecutive values is below epsilon for the last window steps.
has_converged_by_patience(values, patience)	Check if an array has converged by no improvement over a recent window.
has_converged(values, *[, rtol, atol, ...])	Check if an array has converged by various metrics.