PENERAPAN DATA MINING MENGGUNAKAN NAÏVE BAYES UNTUK OPTIMALISASI PENJUALAN SPAREPART MOTOR

This study focuses on the challenges of inventory management and marketing strategies faced by PWT-Part Timika Wholesale Store, which are often hampered by delivery delays. To overcome this, a data mining method using the Naïve Bayes algorithm was applied to optimize motorcycle spare parts sales. The objective of this study is to optimize motorcycle spare part sales by analyzing sales data and classifying spare parts based on their brands. The research approach used is a quantitative method involving the Knowledge Discovery in Database (KDD) process, which involves processing 1,010 sales data points from 2021 to 2023. After preprocessing, 567 data points were used, with 80% (454 data points) allocated for training and 20% (113 data points) for testing using the Weka application. The test results showed that the Naïve Bayes model had good classification performance, with an accuracy of 82.30%, precision of 81.8%, and recall of 82.3%. The conclusion of this study confirms that the Naïve Bayes algorithm is capable of classifying parts based on brand, not just the categories “popular” or “unpopular.” This enables PWT-Part to identify customer preferences more specifically, design more targeted marketing strategies, and optimize inventory.