Demand forecasting for online market stock: Case study cleanroom apparel

Abstract

This research aims to study and develop a forecasting framework for an appropriate production planning demand as well as to analyze the trend of future sales in order to plan the production in line with an increased demand by exploring time series forecasting. This paper studies data characteristics of past volumes of goods sales namely Product A B C E and L, so that an appropriate forecasting technique can be chosen. By comparing 4 forecasting methods including Moving Average, Single Exponential Smoothing, Double Exponential Smoothing, and Regression Analysis Method. Test results show that the forecasting method giving the least errors for Product A is Regression Analysis Method, with the equation Y=403.4-0.62x which gave the lowest MAPE value equals to 22.03. The economic order quantity (EOQ) of Product is 172 units with the total cost of 27,345.51 Baht. Whereas, the forecasting method for Product B is Single Exponential Smoothing Method with a value equals to 0.056, which gave the lowest MAPE value at 72.20. The EOQ of Product B is 150 units with the total cost of 23,280.66 Baht. The forecasting method for Product C is Regression Analysis Method, with the equation Y=417.4-0.82x which gave the lowest MAPE value equals to 28.1. The EOQ of Product C is 193 units with the total cost of 24,953.52 Baht. The forecasting method for Product E is Moving Average N=3, which gave the lowest MAPE value equals to 31.5. The EOQ of Product E is 336 units with the total cost of 14,109.57 Baht. Lastly, the most appropriate forecasting method for Product E is Regression Analysis Method, with the equation Y=1092-3.88x which gave the lowest MAPE value equals to 47. The EOQ of Product E is 1844 units with the total cost of 6,639.97 Baht. © 2019 Association for Computing Machinery.