基于组合模型的云南省卷烟需求预测与结果评价研究

本文提出了一种适合云南烟草行业需求预测的方法体系,构建了优化的回归预测模型和时间序列预测模型,对云南省卷烟需求量进行预测:在数据处理阶段,对价值量指标进行CPI还原;在指标选择阶段,运用时差相关分析和简单相关分析确定宏观经济指标;在模型构建阶段,运用先行两期的宏观经济数据预测当期卷烟销量,并运用最小二乘法计算回归预测模型和时间序列预测模型的权重;在对模型稳定性评价阶段,以建模预测结果的平均相对误差及置信度为基础,构建了一套适合云南卷烟需求预测的模型评判标准,并根据云南实际对模型的稳定性进行评价。 

Prediction of cigarette demand in Yunnan province based on combination model and criteria of result evaluation

In this paper, a system suitable for Yunnan tobacco industry demand forecasting was proposed, and an optimized regression prediction model and time series prediction model were constructed to predict cigarette demand in Yunnan province.In data processing stage, CPI reduction was carried out for value index. In the stage of index selection, time difference correlation analysis and simple correlation analysis were carried out to determine macroeconomic indicators In the model construction stage, the first two periods of macroeconomic data were used to forecast current cigarette sales, and weights of the regression model and time series prediction model were calculated by least squares method. In the evaluation stage of model stability, a set of criteria for evaluating forecast of cigarettes in Yunnan was constructed based on average relative error and confidence of model predictive results, and the stability of the model was evaluated according to actual situation of Yunnan province.