Application of the C4.5 Decision Tree for Classifying Furniture Product Sales Performance at CV Indowest Design Furniture

Many rattan furniture producers, including CV Indowest Design Furniture, still judge product performance manually, so production, restocking, and promotion decisions rest on informal estimates. This study applied the C4.5 decision tree algorithm to classify the company's products by sales performance and embedded the model in a web-based decision support system. Sales records from August to November 2025 (99 transactions, 1,875 units) were aggregated into 17 product-level records. Each product was labelled best seller, moderate, or low seller from its average units per transaction using management-defined thresholds, and purchase frequency and unit selling price, each discretized into three levels, served as predictors. The tree was grown with the gain ratio criterion and implemented in a Laravel 10 and MySQL application. Unit selling price was selected as the root node (gain ratio 0.5585), and the final tree contains five rules. On the 17 training records the tree reached an accuracy of 76.47%, with macro-averaged precision, recall, and F1-score of 77.14%, and all five high-priced products were correctly identified as low sellers. Under leave-one-out cross-validation, however, accuracy fell to 47.06%, only slightly above the 41.18% majority-class baseline, because the two predictors could not reliably separate best sellers from moderate products. The tree therefore offers an interpretable profile of current sales performance, particularly for flagging high-priced slow movers, but it is not yet a dependable predictive tool; longer transaction histories and richer attributes are needed before it is used for forecasting.