Multi-Strategy Pruning System for High-Value Retail Transaction Analysis in Grocery Stores

Abstract

This research proposed multi-strategy algorithms for high-value retail transaction analysis using pruning systems, such as Minimum Significant Utility (MSU), Optimal Support Reduction (OSR), Optimal Weight Learning (OWL), and Cost Complexity Post-Pruning (CCP). The proposed framework was designed to remove negligible patterns and enhance the classification performance. The Gini Index was then used as the splitting criterion to build a Classification and Regression Tree (CART) model. The algorithm was tested on a Multi-Strategy Pruning System for retail transaction analysis in a Grocery Store using the Happy Grocery Dataset, which contains 10,000 transactions and unique items. Traditional mining techniques often yield complex decision trees that contain redundant patterns, irrelevant attributes, and noisy branches, thereby increasing their complexity and incomprehensiveness. Before post-pruning optimization, the suggested pruning procedures are used to eliminate low-utility transactions, weak support patterns, and less significant qualities. Accuracy, precision, recall, F1-score as a metrics to find pruning ratio, and tree complexity used for the experimental evaluation. Intelligent retail analytics applications include recommendation systems, transaction latency, personalized marketing, consumer retention, and strategic business planning.