Penerapan Metode Naïve Bayes untuk Klasifikasi Produk Berdasarkan Kategori Penjualan di Toko Artemist

Abstract

The implementation of this model offers practical benefits for stock management, promotional planning, and data-driven product strategy decisions, thereby improving operational efficiency for medium-scale retail businesses. The application of data analysis in the retail sector is essential to support accurate and efficient decision-making. This study aims to classify products at Artemist Store into two categories: high demand and low demand, using the Naïve Bayes method. The data used are sales records for one year with a total of 8,106 transactions, which after preprocessing resulted in 148 products. Class labels are determined based on the average sales threshold. The dataset is divided using a stratification scheme of 70% training data (103 products) and 30% test data (45 products). The Naïve Bayes algorithm is implemented in RapidMiner Studio software. The evaluation results on the test data show an accuracy of 93.33%, with 89.29% precision and 100% recall in the high demand class, and 100% precision and 85% recall in the low demand class. These findings prove that Naïve Bayes is effective in identifying products with different levels of consumer interest, while also providing practical benefits in the form of stock management recommendations, promotional planning, and data-driven marketing strategies for medium-scale retailers.