Penerapan Metode Naive Bayes untuk Klasifikasi Produk Kurang Diminati Berdasarkan Data Penjualan di Toko Laris Eksis

Abstract

This study aims to apply the Naïve Bayes algorithm to classify in-demand and less in-demand products at Toko Laris Eksis based on sales data, including attributes such as the number of product page views (view), the number of products added to the cart (cart), and the number of products sold (sales). The dataset consists of 245 products from 516 sales transactions after data cleaning. The results show that, despite the class imbalance, the Naïve Bayes algorithm achieved an accuracy of 97.26%, with 100% precision and 96.8% recall for the Less In-Demand class, and 84.6% precision and 100% recall for the In-Demand class. This model outperforms the majority baseline accuracy of 89%. These findings indicate that the Naïve Bayes method is highly effective in detecting in-demand products, even with imbalanced data. Practically, this model can support decisions related to promotions, bundling, and stock clearance in retail. Future research is recommended to use k-fold stratification for evaluation, test adaptive thresholds, and integrate the model into an interactive visual dashboard.