AI-Assisted Forecasting of UNIBA SEHAT Water Demand Using Exponential Triple Smoothing and Weighted Moving Average Methods

Abstrak

This study focuses on the implementation and analysis of demand forecasting methods for bottled mineral water products at PT. ARSINUM, located on Jl. Raya Lenteng No. 10, Batuan, Sumenep - Madura. Currently, the company employs a basic forecasting approach that relies solely on data from the previous period, resulting in low forecasting accuracy and effectiveness. To address this issue, the study evaluates two forecasting methods: Exponential Triple Smoothing (ETS) and the Weighted Moving Average. The objective is to identify the most accurate method to minimize forecast errors and improve operational and supply chain efficiency. The research methodology includes interviews, observations, historical data analysis, and literature review. Based on the findings, the Exponential Smoothing method with a smoothing constant (α) of 0.2 proved to be the most accurate, with a forecasted demand of 3,040 units of bottled mineral water in various packaging sizes for December 2024. This recommendation aims to support PT. ARSINUM in enhancing its demand planning and inventory management processes.