KLASIFIKASI PERFORMA PENJUALAN AKSESORIS MOTOR MENGGUNAKAN ALGORITMA C4.5 DENGAN PENDEKATAN DATA MINING

Abstract

This study aims to develop and evaluate a sales performance classification model for motorcycle accessories using the entropy-based Decision Tree (C4.5) algorithm, with Random Forest employed as a comparative model. The main objective is to identify the transaction-related factors that most significantly influence sales performance (High or Low) in an accurate and interpretable manner, thereby supporting inventory management decisions in motorcycle accessory businesses. The dataset consists of 832 motorcycle accessory sales transactions recorded throughout 2021, including attributes such as unit price, quantity sold, product category, and transaction month. The research methodology comprises data collection, preprocessing using one-hot encoding, stratified training and testing data partitioning (80:20), model development using the C4.5 and Random Forest algorithms, and performance evaluation based on accuracy, confusion matrix, classification report, ROC-AUC, and 5-fold cross-validation. The experimental results show that the C4.5 model achieved a test accuracy of 79.04% with an average cross-validation accuracy of 81.25%, while the Random Forest model achieved an accuracy of 76.05%. Feature importance analysis indicates that Quantity Tier and Unit Price Tier are the two most influential factors in determining sales performance. In conclusion, the C4.5 algorithm is effective for classifying motorcycle accessory sales performance, as it provides competitive predictive accuracy while maintaining high model interpretability, making it suitable for supporting inventory planning and marketing strategy development.