Machine Learning Sales Forecasting for Food Supplements in Pandemic Era

Abstract

The Covid-19 pandemic has brought a lot of concerns about the operational and financial situation of businesses. Forecasting is crucial as it guides businesses through these critical points. Forecasting has become even more critical in the pandemic environment and therefore the necessity of using an accurate forecasting method has increased. Taking this into consideration, in this study, intelligent machine learning methods, namely; Grey Model (GM), Artificial Neural Network (ANN) and Support Vector Machine (SVM) are applied to make a short-term prediction of a food supplement, a product whose demand increased with the pandemic situation. Eighty-five percent of the historical data is used for training purposes and fifteen percent of the data is used for measuring accuracy. The accuracy of the models employed is improved with parameter optimization The accuracy performance indicator Mean Absolute Percentage Error (MAPE) showed that all methods give superior results when the historical data has an increasing sales trend. This study presents an important consideration for businesses and has a potential to be generalized for a business whose sales have an increasing trend not only because of the pandemic but also for any reason.