get_changepoints: Estimate anomaly-score for each value of each Key of a data...

In td-berlin/anomalizer: Implementation of Outlier-Detection-Methods.

Description

Estimate anomaly-score for each value of each Key of a data frame in three steps

Usage

get_changepoints(
  x,
  min_segment_size = 50,
  min_segment_share = 0.2,
  min_RSS_decrease = 0.2
)

Arguments

x	vector containing time-series values
min_segment_size	minimum number of datapoints between each changepoint. Only taken if > min_segment_share*N
min_segment_share	minimum share of datapoints between each changepoint. Decreasing this number fucks performance exponentially
min_RSS_decrease	minimum RSS decrease to apply changepoint. The lower, the more changepoints are found

Value

Boolean-Vector of size x saying if realted value is a changepoint

td-berlin/anomalizer documentation built on Feb. 21, 2020, 2:03 a.m.