Algorithms for change point detection

[johnaohara]

Currently Horreum implements 2 different change dectections alorithms, "Fixed Threshold" and "Relative Difference of Means". Whilst these 2 provide a solid basis for a lot of change detection use cases, there are others that might be a better fit for automated change detection.

The aim for introducing/creating new algortihms would be to lower the rate of false positives and false negatives and explore if non-parametric algorithms perform better than parametric algorithms where the user needs to tune the change detection algorithms to reduce the rate of false +ve/-ve

Some existing examples / research exists;

• https://arxiv.org/pdf/2301.03034.pdf
• Change detection improvements #38

it would be interesting to evaluate either existing or novel approaches to determine what approach provides the best outcome against existing datasets