Optimasi Peramalan Penjualan Perishable Product dengan Metode Time Series Menggunakan Software POM-QM

Abstract

Perishable products are items with short shelf life. In this study, perishable product being analyzed has only one day shelf life. Therefore, an accurate forecast method is needed to avoid shortage and waste. Shortage occurs whend demand exceeds the company’s stock, resulting potential loss sales. Waste occurs when demand is lower than available stock, leading to overstock. Excexx inventory causing losses and reduces profit. This study is conducted using 2 scenarios, as shown in Table 1. Scenario 1 uses 60 days od data, while scenario 2 separates data between weekend and weekdays. Data abalysis was performed using time series method with POM-QM software. The best method is determined based on lower MAD, MSE, MAPE. The result of this study sho that separating the historical data between weekends and weekdays (scenario 2) leads to better forecasting accuracy, indicated by a lower MAPE compared to scenario 1. For weekend forecasting, the best method is exponential smoothing (Scenario 2A.2) with MAPE 10,05%. For weekday forecasting, the best method is moving average (Scenario 2B.1) with MAPE 15,40%. The research is expected to serve as a reference for company in selecting appropriate forecasting method. The goal is to help company anticipate shortage and waste in perishable products. The recommendation for company is automate the separation of weekend and weekday data to accelerate the forecasting prosess. Future research recommendation is use other forecast method and software, then consider factor consists of shelf life, promotion, national holidays, or national event.