Research on Short-Term Forecast of Product Demand Based on SARIMAX Model

As the external environment of enterprises is constantly changing, the accuracy of demand forecasting, as the first line of defense of enterprise supply chain, is generally low due to the influence of many factors. Based on product order data, this paper analyzes the influence of different prices, different sales methods, different time periods and other factors on product demand, and establishes a SARIMAX model for demand prediction, which provides decision-making basis for enterprises to rationally schedule commodity inventory and formulate marketing strategies. Firstly, the data is preprocessed, and the correlation analysis is made between various factors and the product demand, and the characteristics and ways of the influence of the factors with greater correlation on the product demand are analyzed. The short-term forecast model of product demand based on SARIMAX is established, and the forecast is carried out according to different time granularity such as day, week and month. After determining the optimal parameters of the model, the cleaned and processed data are input into the established SARIMAX model, and a variety of evaluation indexes are used to measure the prediction effect. Next, the influence of different time granularity on the prediction accuracy was explored. Under the time granularity of day, week and month, the prediction accuracy of SARIMAX model was 95.86%, 93.11% and 95.62%, respectively. It can be seen that the prediction accuracy based on day time granularity is the highest, followed by month time granularity. Three kinds of time granularity from large to small prediction accuracy are day, month and week respectively. In this paper, the time series decomposition model is used to forecast the product demand in the short term, and the validity and accuracy of the model are evaluated. The results show that the model can be effectively applied to the actual product demand forecast.