KPI Request: No. of correlated test failures

[Shreyanand]

As an Openshift developer or manager, I would like to know for each test, the average number of other tests that are most likely to fail during the same build. If a test has a lot of failed correlated tests, it indicates that there could be issues with this test as well and vice versa.

@MichaelClifford already wrote a notebook that finds correlation between tests. This notebook can be repurposed for this metric.

Acceptance criteria

• small notebook in notebooks/data-sources/TestGrid/metrics/ that collects this metric and stores it in ceph as a parquet