中国福利彩票销售额影响因素分析与基于残差主成分分析的预测

本文应用多元回归分析的方法讨论了中国各省份福利彩票销售额与国民经济、就业人员和职工工资、财政、固定资产投资、人民生活这五个大类中的22个预测变量之间的相关关系,找出了对彩票销售额有显著性影响的变量,此外我们发现,不同年份的数据所得到的残差存在强相关性,本文利用主成分分析的观点提取出了不同年份残差中的公共信息,最后得出彩票销售额的预测方程,利用上述方程对2007年各省份的彩票销售额进行预测,并与真实的销售额数据进行对比,获得了较小的平均相对预报误差.这说明本文所建立的模型确实可以对中国的福利彩票销售额进行有效的预测.

The Acting Factor Analysis and the Prediction Based on the Principal Component Analysis of the Residual of Chinese Welfare Lottery Sales

By introducing the approach of multivariate regression analysis,the relationship between Chinese welfare lottery sales and 22 social and macro economic predictor variables was discussed and the crucial predictor variables,which remarkably influence the lottery sales,were selected.Moreover,we found that there was strong dependence among the residual which was obtained by fitting the data of different years.Using the principal component analysis,the common information of the residual among different years was extracted and the prediction equation was obtained.Using the equation suggested by the author,the lottery sales in year 2007 were predicted.The similarity between the fitted values and the real data in year 2007,which was measured by the average percentage of predicted error,fully reveals that our model may be effective for the real application of the prediction of lottery sales in China.