Analysis of material demand prediction using time series method to minimize PT XYZ Redpost data

Abstract

Inaccurate inventory management can lead to stockouts (RedPost) which disrupts the production process in manufacturing companies. This study aims to analyze material demand patterns and determine the most accurate forecasting method to minimize RedPost occurrences at PT XYZ. The method used is a quantitative approach with time series forecasting techniques, namely Double Exponential Smoothing (DES), Moving Average, and linear regression, using historical demand data for one year obtained through documentation and interviews. The results show that the Double Exponential Smoothing (DES) method provides the best level of accuracy with the lowest error value, namely MSE of 5312.426 compared to other methods. The characteristics of demand data that are non-stationary and tend to follow trend patterns make trend-based methods more appropriate to use than simple average methods. Thus, the application of the Double Exponential Smoothing (DES) method can help to reduce the risk of RedPost and improve efficiency and inventory management, as well as support more effective, measurable, and data-based operational decision making.