Data analytics system for product management in retail business

Abstract

This paper aims to develop a data analytics system for product management in retail businesses based on the Winters' Method and tests the milk sales forecasting model with machine learning using linear regression, random forest, and eXtreme Gradient Boosting (XGBoost). Using Minitab, apply machine learning test data to separate milk statistical analysis with five forecasting methods. Actual sales figures were compared to the difference using the results of the machine learning model and statistical analysis. The results of machine learning show that random forests have more minor deviations than a statistical analysis performed with Minitab that showed Winters' Method (Additive Method) with UHT of less than 7.9 percent. Pasteurized was less than 9.54 percent. Soymilk was less than 12.57 percent. Fermented milk was less than 10.47 percent. The result is a desirable user satisfaction rating of 4.47 out of 5. © 2022 IEEE.