AUTOMATION AND OPTIMIZATION OF SOFTWARE CONFIGURATION MANAGEMENT PROCESSES USING MACHINE LEARNING TECHNIQUES
Keywords:
Machine Learning (ML), Software Configuration Management (SCM), Optimization, Configuration Audit (OCA), Normalization, Decision TreesAbstract
Software Configuration Management (SCM) plays a critical role in modern DevOps pipelines by ensuring
consistency, integrity, and control over software builds and deployments. This study presents a machine learning
based framework to automate and optimize SCM processes. By leveraging historical SCM metrics and deployment
success labels, we trained and evaluated a Random Forest classifier that achieved over 92% accuracy. Feature
importance analysis revealed that metrics such as code churn, task complexity, and structural cohesion strongly
influence deployment outcomes. A Streamlit-based dashboard was also developed to visualize predictions and
performance metrics. The proposed solution demonstrates the practical viability of applying machine learning to
enhance decision-making, reduce manual intervention, and improve deployment reliability in continuous
integration/continuous deployment (CI/CD) environments.
