How Your Test Suite Can Predict Production Failures with Anomaly Detection
from Software Testing with Fexingo: QA, Automation, and Reliable Software Engineering Hosted by Fexingo
Episode 86 of Software Testing with Fexingo: Lucas and Luna dive into a powerful technique that goes beyond traditional pass/fail assertions: using anomaly detection on test metrics to predict production incidents before they happen. They explore a real-world case where a mid-size e-commerce company used this approach to catch a silent memory leak that standard unit tests missed for weeks. Lucas explains how teams can collect key metrics like response time percentiles, error rates, and resource utilization from their test runs, then apply simple statistical models—like moving averages or standard deviation thresholds—to flag deviations. Luna pushes back on the false positive problem, and Lucas shares how rolling baselines and seasonal adjustment kept the alert noise manageable. The episode also touches on tooling: open-source libraries like PyOD and commercial options like Datadog. Listeners walk away with a concrete playbook for adding a 'prediction layer' to their existing test infrastructure. #SoftwareTesting #QA #AnomalyDetection #ProductionPredictions #TestMetrics #Monitoring #PyOD #DataDog #MachineLearning #MemoryLeak #Ecommerce #DevOps #CI_CD #Reliability #FexingoBusiness #BusinessPodcast #Technology #TestAutomation Keep every episode free: buymeacoffee.com/fexingo
Embed this episode
No transcript yet.