PREDICTIVE ANALYTICS FOR PIPELINE FAILURE FORECASTING IN CLOUD DEVOPS ENVIRONMENTS
Abstract
As DevOps environments grow in complexity, identifying potential failures in CI/CD pipelines becomes critical. This study proposes a machine learning-based predictive analytics model that uses pipeline metadata, historical logs, and build performance metrics to forecast potential pipeline failures. Implemented within Azure DevOps and GitLab CI pipelines, the model achieved over 87% accuracy in predicting failures before execution. The system proactively alerts teams and triggers remediation scripts via Python-based automation. This research provides a novel intersection between AI and DevOps, enhancing resilience and reducing downtime in modern software delivery pipelines.