deequ icon indicating copy to clipboard operation
deequ copied to clipboard

Histogram anomaly detector

Open slgobinath opened this issue 4 years ago • 1 comments

Hi, I'm trying to check if the histogram of a column is changing between two versions of the data. Though there is a Histogram analyzer, Deequ's API to use an analyzer for anomaly detection does not support the Histogram analyzer.

The problem comes with the addAnomaly method expecting a Metric[Double] analyzer. It would be great if the API is modified to support Histogram analyzer as an anomaly detector.

slgobinath avatar May 12 '21 14:05 slgobinath

We are open to developments in this direction. The implementation will be tricky as the anomaly detection needs to be adopted to accomodate the new kind of metrics. We currently don't have the bandwidth to implement this ourselves but would be happy to review a (concise) PR on this.

twollnik avatar Jul 20 '21 15:07 twollnik