popmon

Why we included this project

Most drift-detection tools expect you to stand up an entire monitoring stack before you learn anything useful. popmon takes the opposite route: it runs on existing pandas or Spark dataframes, buckets feature values over time, computes histograms, and applies statistical tests to catch trends, shifts, peaks, or correlations that have changed. The output is a self-contained HTML report, so findings are easy to share or file away. Because the same logic works on a local dataframe and a distributed one, teams can prototype checks in a notebook and reuse them on cluster pipelines. You can also set business rules to flag problems automatically and ship histogram data to Grafana or Kibana if your team already lives in those dashboards.