Predictive analytics and machine learning techniques for enhanced cloud computing resource management

This study explores the application of machine learning (ML) and predictive analytics for optimizing cloud resource management and capacity planning, addressing the limitations of traditional static approaches. Using a systematic literature review under PRISMA guidelines, the research analyzed over 4,800 studies from major academic databases, ultimately selecting 42 high-quality papers for detailed evaluation.

The study compared multiple ML models — including Random Forests, Neural Networks, Linear Regression, and Polynomial Regression — using performance metrics such as Mean Squared Error (MSE), Mean Absolute Error (MAE), and R² scores. Results showed that Random Forest algorithms consistently outperformed traditional methods, achieving over 85% accuracy in predicting cloud resource utilization, particularly for storage and memory. Neural networks and regression models showed variable performance across different resource types, while CPU utilization remained the most complex to predict.

The findings highlight that ML-driven dynamic resource allocation enables more proactive scaling, cost efficiency, and performance optimization compared to static models. Successful implementation depends on data quality, model complexity, and real-time processing capabilities. Overall, the study concludes that machine learning and predictive analytics are practical and effective tools for enhancing cloud infrastructure efficiency, cost reduction, and service reliability.