Research on Precise Demand Prediction of New Retail Target Product Based on Dual Model

Abstract

In this study, the target data set is firstly obtained by data processing, and then the Grey Verhulst model and ARIMA model are respectively used for modeling and prediction research on the target data set, so as to calculate the predicted value. Then MAPE (average absolute percentage error) is calculated according to the predicted value. The typical characteristics and adaptability of Grey Verhulst model and ARIMA model are compared and analyzed. This study shows that the ARIMA model is more accurate than the Grey Verhulst model in the short term, and its prediction accuracy decreases sharply with the extension of time, which is suitable for the short term prediction. The accuracy of Grey Verhulst model is relatively stable, and the accuracy is improved with the extension of time, which is suitable for medium and long term prediction.