关键词: 自回归滑动平均模型, 修正因子, 预测, 前处理

Abstract: In order to improve the accuracy of sales forecast to provide references for business decisions, a forecasting model for sales is set up to forecast the monthly total sales based on ARMA (autoregressive moving average model). The weight of each affecting factor of the input time series is adjusted by adopting the modifying factors (pretreatment). Then, the ARMA model is built. Finally, the forecasting results are re-modified (post process) which improves the accuracy of sales forecast. By means of the IIS as the application server and the Oracle database as the database server, the B/S architecture and the design of ASP.NET four-layer architecture are adopted. The forecasting system includes functions like modification of the input sales data, model recognition, order selection, parameter estimation, modification of forecasting results and interface display to implement the product sales forecasting system.


Key words: ARMA model, correction factor, forecast, pretreatment