Clustering Failures by Root Cause

Use when facing many failing tests, false positives, review findings, or bug reports at once, or when improving a metric like precision or recall without regressing another.