Prediksi Permintaan Stok dan Tingkat Pendapatan Menggunakan Decision Tree Learning: Studi Kasus Ponti Kreasi Utama

Rising market saturation leads to a need for new adaptation strategies, such as enhanced customer service through faster and more consistent order fulfillment time. This depends on efficient inventory stock management, ensuring the availability of necessary materials. This research aims to leverage the local business in the Asia-Pacific region with the insight required for inventory stock management and purchase planning. This research uses a quantitative approach utilizing decision tree learning to analyze 71,000 samples of historical transaction data between January 1, 2018, and December 31, 2023, from PT Ponti Kreasi Utama, a family printing business located in Pontianak, Indonesia. The goal is to predict highest-selling materials, resulting in an error value of 0% to 40% depending on timeframe length. Monthly material usage quantities are also predicted with an error value of 73.8%; upon further analysis the error value is effectively halved in practice. This prediction allows the business to proactively restock in anticipation of demand, leading to improved customer service and less occurrences of order fulfillment delays and cancellation. Future research could explore a hybrid approach combining multiple machine learning techniques for a further increase of predictionprecision.