Double exponential smoothing brown method towards sales forecasting system with a linear and non-stationary data trend

Abstract

UD Parama Store is a trading company engaged in selling retail goods that sell various types of daily necessities retail. The major problems occurred are the difficulty in predicting the sales due to the maturity level of experience, customer demand changes, and the owner's limited memory. Therefore, there should be an increase of merchandise stock to prevent any sudden decrease in sales and overcome the stock shortage when there is an increase in sales. In this current research, a sales forecasting web-based system was designed and built to assist the owner in predicting the number of sales in the next period. As a result, decisions can be made in determining the number of goods to be provided. The forecasting method used was double exponential smoothing brown by improving forecasting, averaging (smoothing) the past value of a time coherent data, and decreasing (exponential), which requires one parameter only. It was used to increase and decrease the linear and non-stationary data. The calculation of forecasting accuracy using the MAPE (Mean Absolute Percentage Error) method in forecasting sales of merchandise produces the smallest error rate ranging from 7.99% to 32.42% for 10 different items.