#time-series #detection #time-series-analysis #bayesian #online #point #changepoint

augurs-changepoint

Changepoint detection using Bayesian Online Change Point Detection and more

21 releases (8 breaking)

0.9.0 Jan 14, 2025
0.8.0 Dec 23, 2024
0.7.0 Nov 25, 2024
0.3.1 Jul 30, 2024

#416 in Machine learning

Download history 762/week @ 2024-10-16 475/week @ 2024-10-23 70/week @ 2024-10-30 349/week @ 2024-11-06 47/week @ 2024-11-13 313/week @ 2024-11-20 44/week @ 2024-11-27 13/week @ 2024-12-04 20/week @ 2024-12-11 120/week @ 2024-12-18 30/week @ 2024-12-25 50/week @ 2025-01-01 120/week @ 2025-01-08 69/week @ 2025-01-15 1/week @ 2025-01-22 11/week @ 2025-01-29

249 downloads per month
Used in augurs

MIT/Apache

47KB
520 lines

Changepoint detection models

Changepoint detection of time series.

For now it is mostly just a wrapper around the changepoint crate, with a common Detector trait to allow for more implementations in future.

Example

use augurs::changepoint::{Detector, DefaultArgpcpDetector};

let data = [0.5, 1.0, 0.4, 0.8, 1.5, 0.9, 0.6, 25.3, 20.4, 27.3, 30.0];
let changepoints = DefaultArgpcpDetector::default().detect_changepoints(&data);
// 0 is always included. 6 is the index prior to the changepoint.
assert_eq!(changepoints, vec![0, 6]);

Credits

The bulk of the actual work is done by the changepoint crate.

Dependencies

~16MB
~285K SLoC