Abstract

Cloud computing has become a cornerstone for machine learning (ML) applications, offering scalable infrastructure to process vast amounts of data. This study evaluates SQL-driven and Python-driven workflows in cloud-based ML, focusing on execution time, cost efficiency, and performance across platforms like AWS, GCP, and Azure. Results reveal that SQL-driven workflows excel in speed and cost-effectiveness for structured data preprocessing, while Python-driven workflows provide superior flexibility and accuracy for advanced analytics and modeling. A hybrid approach integrating both workflows is recommended to optimize resource utilization and achieve a balance between efficiency and performance. The findings underscore the importance of selecting appropriate cloud resources and adopting monitoring tools to ensure scalability and cost control. These insights provide a roadmap for organizations seeking to enhance the efficiency and effectiveness of their cloud-based ML operations