Skip to content

Add Weighted RMSE Score - #36

Merged
roelofsc merged 3 commits into
mainfrom
develop
Oct 6, 2026
Merged

roelofsc merged 3 commits into
mainfrom
develop

Conversation

@roelofsc

@roelofsc roelofsc commented Oct 6, 2026 •

Copy link
Copy Markdown
Contributor

Adds PR #34

Adds a new WeightedRMSEScore anomaly score (weighted_rmse) that applies per-feature weights to standardized construction errors before computing the RMSE, enabling feature-specific importance in anomaly detection.

Removes sync to gitlab workflow (not needed and does not work anymore).

cgueck895 and others added 3 commits October 6, 2026 08:33
* Add weighted RMSE anomaly score and deprecate RMSEScore scale param

* depracted scale parameter for mahalanobis score

* Added additional full dimensional mahalanobis norm computation and deprecated pca-based mahalanobis_score. Also added documentation on anomaly scores

* Fix WeightedRMSEScore load by giving feature_weights a default value

Loading a FaultDetector reconstructs each sub-model with no constructor
arguments, which previously raised a TypeError for the weighted_rmse score
because feature_weights was a required parameter. Default it to None and
add unit tests covering the FaultDetector save/load round-trip.

* Fixed weighted score by removing second scaling

* Fixed weighted score bug and applied suggestions from review

* Removed and edited comments in weighted_rmse_score
@roelofsc
roelofsc requested a review from cgueck895 October 6, 2026 06:54
@roelofsc
roelofsc merged commit ace3424 into main Oct 6, 2026
3 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants