Integrating Data Science and Meerkat Clan Optimization for Predictive Cloud Task Scheduling

An intelligent predictive task scheduling framework for cloud computing is presented that integrates data science techniques with the Meerkat Clan Algorithm (MCA) to enhance scheduling efficiency, reliability, and scalability. The framework employs predictive modelling to estimate task execution time, virtual machine (VM) failure probability, and expected throughput, enabling proactive and informed scheduling decisions in dynamic cloud environments. These predictive insights are incorporated into an MCA-based multiobjective optimization strategy that simultaneously minimizes makespan and VM failure rate while maximizing throughput. The effectiveness of the proposed approach is evaluated using realistic Alibaba GPU workload traces and implemented within the SimPy discrete-event simulation environment to accurately capture cloud system dynamics. Comparative experimental results against Ant Colony Optimization (ACO), Deep Reinforcement Learning (DRL), and Enterprise Development Optimizer Algorithm (EEOA) demonstrate substantial performance improvements. On average, the proposed scheduler achieves makespan reductions of approximately$\mathbf{5 5. 4 \%}$, lowers VM failure rate by$\mathbf{5 1. 5 \%}$, and improves throughput by 34.4 % across varying workload scales. These results confirm that combining predictive analytics with clanbased meta-heuristic optimization provides a robust and scalable solution for next-generation, data-intensive cloud computing task scheduling.