An Efficient Load Balancing Algorithm for Mining Frequent Item sets for Large Transactional Data

Abstract

Frequent Itemset Mining (FIM) is critical in many data-intensive applications such as customer behavior analysis, recommendation systems, and fraud detection. Nevertheless, while the size and complexity of transactional databases increase explosively, classical FIM methods face tremendous challenges in scalability, processing efficiency, and system reliability. The growing amount of data can result in loaded servers and virtual machines (VMs), which cause performance reduction, signal loss, and longer execution time. Such problems not just hinder system stability but also lead to a bad user experience. This research proposes a new approach known as Grey Wolf-based Adaptive Frequent Itemset Mining (GW-AFIM), combining intelligent load balancing with frequent itemset mining to address current limitations. The model applies Grey Wolf Optimization (GWO) to automatically evaluate and assign data loads between processing units. It draws inspiration from the manner in which grey wolves hunt and live in packs. The fitness function of the GWO algorithm is responsible for minimizing data traffic, enhancing load balancing, and accelerating computation.GW-AFIM aims at mining transaction-based frequently preferred itemsets with balanced utilization of resources. Rigorous experimental analyses prove that GW-AFIM is remarkably better than current algorithms with respect to processing time, speedup, accuracy, scalability, and error rate. The outcomes show that the proposed model is an efficient and scalable solution for mining frequent itemsets from enormous transactional data, which is very much applicable in contemporary big data environments where real-time analytics and fault tolerance are of utmost importance.