whylogs
Open-source data logging for ML pipelines with drift and data-quality checks.
About
whylogs is an open-source library from WhyLabs for logging statistical summaries of datasets, called whylogs profiles. Data scientists and ML engineers use it to detect data drift in model input features, catch training-serving skew, validate data quality in pipelines, and track data distributions over time.
The library supports privacy-preserving data collection and ships with data constraints that alert when data stops looking the way it should. It installs with a single pip command and plugs into the WhyLabs observability platform for hosted monitoring.
What it can do
- Privacy-preserving statistical profiles of any dataset
- Data drift, training-serving skew, and concept drift detection
- Data constraints and quality validation for ML pipelines
Setup
Install with pip install whylogs, then call why.log on a pandas dataframe or dataset.
Prerequisites
Python environment and ML data or model inputs to profile.
Use cases
Works with
Trust score — 7/10
Scored 7/10 (metrics tier, metrics-v1). Evidence: Verified via vendor site/docs/GitHub and reputable coverage, wave 103; Trending score 74/100 from researcher assessment.
★ 3,400 GitHub stars
Sources
Every fact above was checked against these sources. Where a claim comes from the provider and couldn't be independently verified, the text says so.