Demand Forecasting UMKM Kopi Keliling Berbasis Deep Learning Klasik

Hanifah Muthiah (Sekolah Tinggi Ilmu Ekonomi Bima)
Amirulmukminin (Sekolah Tinggi Ilmu Ekonomi Bima)

Jurnal Manajemen, Bisnis dan Kewirausahaan
Vol. 5 No. 3 (2025): Desember : Jurnal Manajemen, Bisnis dan Kewirausahaan

Mobile coffee MSMEs are part of the creative economy sector that is rapidly growing in urban areas. However, these businesses face uncertainty in daily demand, which is influenced by time, weather, location, and consumer trends. Accurate demand prediction is required to optimize inventory management, reduce the risk of losses, and increase profitability. This study aims to apply a classical deep learning approach, namely Long Short-Term Memory (LSTM), to predict the daily demand of mobile coffee MSMEs. The research data includes daily sales over 18 months with external variables such as weather, weekdays/holidays, and location. The research results indicate that the LSTM model is able to capture seasonal patterns and trends better than classical methods (ARIMA), with higher accuracy for the 7-14 days prediction horizon. These findings support data-driven decision-making for MSME actors in managing inventory, determining strategic sales locations, and designing effective promotions.