With the rapid development of China’s tobacco economy, extent size of market demand for cigarettes determines the direction of the company’s decision-making,so the sales forecasting of cigarette is very important for the market competitiveness of China’s tobacco industry.On the one hand, because of the sales forecasting of cigarette is a time series prediction, so the paper introduces the knowledge about the time series prediction. More than that, cigarette sales have characteristics on non-linear, and coupled with the limitations of traditional statistical forecasting methods and neural network can reflect the characteristics of the nonlinear, the article be elaborated neural network theory. After that, I using clustering method to take a reasonable division on the company’s customers in the business lines and operating volume. Secondly, in order to be able to more performance the characteristics of cigarette sales in the long-term trends and short-term fluctuations, I established TBP(Trend Back Propagation) model and PBP(Period Back Propagation) model based on BP(Back Propagation) neural network according to the theoretical knowledge of the time series and neural network. Similarly, I established TRBF(Trend Radial Basis Function) model and PRBF(Period Radial Basis Function) model based on RBF(Radial Basis Function) neural network.Finally, I investigate the cigarette sales forecast model.On the other hand, found in the companies operating status of the problems and deficiencies, I accord to the company’s historical data and the status worked out a precision marketing program of work. And make a summary and outlook for the next work. |