Data Drift Check

Tabular Data Drift Detector & Reporter

Tabular Data Drift Detector & Reporter:
A CLI tool that connects to any database or CSV, computes statistical drift (KS-test, Jensen-Shannon) between “baseline” vs. “current” data, and emits a Markdown/HTML report with charts.

What it is


A CLI tool that connects to any database or CSV, computes statistical drift (KS-test, Jensen-Shannon) between “baseline” vs. “current” data, and emits a Markdown/HTML report with charts.

Why you need it

• MLOps gap: You need data-drift checks in production, but frameworks like Evidently or Great Expectations are heavyweight.
• Teams want a bullet-proof, “run-me-with-one-command” script for nightly checks.

Key features

python drift-check.py
--baseline base.csv
--current new.csv
--report drift_report.md

• Auto-detects numeric vs. categorical
• Flags features with > 0.1 JS divergence
• Generates simple Matplotlib histograms inline

Installation

pip install pandas numpy scipy matplotlib

Quickstart

python driftcheck.py \
  --baseline base.csv \
  --current new.csv \
  --report drift_report.md