A Hybrid Sales Forecasting Method Based on Stable Seasonal Pattern Models and BPNN

Since many operations of corporation are depend on sales forecasting. It's very important for corporations predict their sales data accurately and reliably. Most of retail businesses are seasonally, therefore, lots of them are using seasonal sales forecasting methods to forecast their sales data and get better results. This paper proposes a hybrid sales forecasting method which combines stable seasonal pattern model and back propagation neural network (BPNN) to forecast retail sales. Practical forecast result in commodity shows it's more accurate than typical seasonal forecasting methods.