Web Based Smartphone Recommendation System Using C4.5 Decision Tree Algorithm for User Need Classification

Abstract

The rapid growth of the smartphone market has created significant difficulty for retail staff in providing consistent, objective purchase recommendations, particularly for customers with limited technical knowledge. This study presents SmartPick, a recommendation system that applies the C4.5 Decision Tree algorithm to address the inefficiency of manual recommendation at Sahabat Ponsel Tembilahan retail store. Purpose: To classify users into smartphone usage categories Gaming, Photography/Videography, Student, and General User based on budget, usage intent, and technical preferences, replacing subjective salesperson judgment with a consistent, rule-based decision process. Methods: A balanced dataset of 500 smartphones was used to train and evaluate the C4.5 model, chosen for its interpretability and ability to represent decision logic as human-readable rules. Results: The model achieved 96.00% testing accuracy, outperforming CART, Random Forest, Naive Bayes, and K-NN. Novelty: Unlike prior recommendation approaches relying on collaborative or content-based filtering, which require extensive historical transaction data, SmartPick integrates technical specifications and budget constraints directly into a single interpretable Decision Tree model, offering a transparent and scalable alternative for retail-based smartphone recommendation that remains explainable to non-technical users.