<![CDATA[Cloud computing has revolutionized information technology (IT) infrastructure by enabling on-demand access to scalable resources. However, the elasticity and complexity of cloud billing models introduce significant challenges for effective resource cost management. This paper proposes a hybrid framework integrating statistical models auto regressive integrated moving average (ARIMA), machine learning techniques long short-term memory (LSTM), and optimization methods deep deterministic policy gradient (DDPG) to forecast and manage cloud costs with enhanced accuracy and adaptability. The framework is empirically validated using synthetic billing datasets and real-world cloud provider data, with performance evaluated via root mean square error (RMSE) and mean absolute percentage error (MAPE) metrics. Results demonstrate 15-25% improvement in cost prediction accuracy over baseline models and up to 20% cost savings through dynamic resource allocation. The framework extends beyond traditional VM-based workloads to support serverless computing (amazon web services (AWS) Lambda and Azure Functions) and container-based applications (Docker and Kubernetes), addressing the growing adoption of microservices architectures. Comparative analysis with existing tools (AWS Cost Explorer and Azure Advisor) reveals superior adaptability in multi-cloud environments. The paper concludes with discussions of emerging paradigms including FinOps practices, AIOps automation, and sustainability-aware resource allocation, outlining future research directions toward explainable AI-driven cost governance.]]>
Copyrights © 2026
