AI engineering glossary · MLOps, LLMOps cloud
What is PSI (population stability index)?
Population stability index (also called PSI): A score comparing the share of data falling into each bin for two datasets, usually training and live data, to measure distribution shift. Practitioners commonly treat values under about 0.1 as stable and above about 0.25 as a significant shift. Formula: PSI = sum of (actual% - expected%) x ln(actual% / expected%).
Population stability index: an example
PSI on the applicant income feature reaches 0.31, so the team investigates and considers retraining.
Related terms
- Data drift
- Kolmogorov-Smirnov test
- Model monitoring
- Feature distribution
More MLOps, LLMOps cloud terms
- Postmortem
- Prometheus
- ReplicaSet
- Resource requests and limits
- Role-based access control
- Rolling update
- Scale to zero
- Secrets management
Where you learn it
Population stability index is part of the AI Engineering Program, which teaches it with notes, a hands-on lab and an exam. The first lesson is free.