Research And Application Of Cigarette Demand Forecasting Model Based Time Series

China tobacco is the state-owned enterprises, the central enterprises, Chinese fiscal revenue and promote local economic development has played a positive role. Chinese economy has entered a "new normal", as a special tobacco industry is facing the plight of downward pressure on the larger economy, foreign invasion, cigarette tobacco control efforts to increase gradually. China tobacco industry is the implementation of vertical management, the state monopoly, according to the order of the system supply of goods. In order to improve the operation efficiency of tobacco enterprises, in this paper the automatic cigarette demand forecasting method, to effectively reduce tobacco industrial enterprises(plant) and commercial enterprises(tobacco companies) inventory, more reasonable arrangements for the production plan to provide reliable technical support. In determining the national tobacco industry management system, Municipal Tobacco Monopoly Bureau(company) the demand forecast "nexus", has important significance. In this paper, cigarette sales history data of SQ city in Henan province for the data set, prediction method of exponential smoothing and seasonal trend with time series, study the prediction model of cigarette demand, and applies the model to the city's monthly cigarette and key brand brand cigarette demand forecast.Based on the above analysis and thinking, this paper makes a concrete research and analysis. First, Analysis of the status quo of the development of China's tobacco industry, the situation faced and the significance of the prediction of cigarette demand, through analysis of the status quo prediction of cigarette demand in the city of SQ, aiming at the existing problems of the artificial method to predict the recommendations for improvement, analysis of the status quo of application of automatic cigarette demand forecast. Then, this paper introduces the definition of cigarette demand forecast, the two main methods of qualitative and quantitative prediction and its specific steps, models are described and studied. Then, the collected data of SQ cigarette sales were collected, combined with the national tobacco policy, analyzed the main influencing factors and the characteristics of cigarette sales. Finally, according to the national tobacco policy and SQ, the results of data analysis, the cigarette sales history data in SQ city as a sample, using three exponential smoothing prediction model and the traditional seasonal trend forecasting model of cigarette demand forecast, and carries on the analysis and comparison of prediction results is established based on the demand of cigarette combination forecasting model TS then, taking Matlab as the development language, the design and implementation of the prediction model of TS system. Using this system, the training data of the cigarette sales data of SQ city from 2010 to 2014 were used to train the training data, and the demand of cigarette in 2015 was predicted. The experiment shows that the average error between the forecast demand and the actual sales volume is 2.07%, which is much lower than the average error of the artificial demand forecasting in the SQ City.