基于组合预测方法的某市香烟销售量预测

A Forecast of Cigarette Sales Based on Combined Forecasting Method

在对某市2010年1月—2015年9月香烟销售量数据分析的基础上,分别建立季节指数预测模型、指数平滑模型和ARIMA模型,样本期内预测对数误差平方和分别为0.9516、1.1474和1.0974。为了提高预测模型的精度,在此基础上建立加权几何平均组合预测模型,样本期内预测对数误差平方和为0.8042。并用组合预测模型对该市所辖区、县香烟销售量总量和主要香烟品牌的需求量预测。

On the basis of the analysis of cigarette sales volume from January 2010 to September 2015 in a city, seasonal index measurement model, exponential smoothness model and ARIMA model were respectively built in this thesis. In the sample period, the quadratic sum of predicted logarithmic errors was 0. 9516, 1. 1474 and 1. 0974. In order to improve the precision of the predicted models, weighted geometric average combined prediction model was set up. The quadratic sum of predicted logarithmic errors in the sample period was 0. 8042. Moreover, the combined prediction model was used to predict the total sales volume of cigarette and quantity demanded of major cigarette brands in districts and counties under the jurisdiction of the city.