A Comparative Analysis of Two Soft Computing Methods for Sales Forecasting in Dairy Production: A Case Study

Abstract

Sales forecasting is a trending area of research in diary production as it helps the industry project the future demand, revenue generation and profit expectation of the industry. Several research methods have been employed in sales forecasting while the Artificial Intelligence (AI) methods offer more plausible and accurate prediction as compared to traditional statistical methods. A new soft computing method, established on the fundamentals of Gene Expression Programming (GEP) is presented in this work to forecast sales in a real-case dairy production company. The study employed back propagation neural network (BNN) which is useful in artificial neural network (ANN) methods. The study also conducted results validation for results’ accuracy, dependability and reliability as well as compared ANN results of prediction against GEP results using R-square and mean square error (MSE). The associated results showed that both AI techniques are robust in sales forecasting. However, the accuracy of GEP was observed to be more promising in comparison to the ANN.