中国农村CPI预测研究

Study on the Prediction of Rural CPI in China

摘要: 本文研究了农村CPI各分类价格指数在1998~2018年变化趋势，对总分类与各分类价格指数进行逐步回归，构建因变量与自变量的多元线性回归关系模型，运用逐步回归方法建立MLR模型，得出影响农村CPI的分类价格指数，确定其相关性变量，为进一步优化相关性系数使模型预测效果更吻合，进一步构建GM(1,N)模型进行预测；再对数据进行光滑性处理，在分析原模型背景值构造形式的基础上对其进行优化，重新构建GM(1,N)模型的背景值，提出了改进后的灰色GM(1,N)模型，通过运算和模型对比分析，发现该改进模型能够有效地提高预测精度，在对农村CPI预测中具有较强的实用性。

Abstract: This paper studies the change trend of each classified price index of rural CPI from 1998 to 2018, makes a stepwise regression between the general classification and each classified price index, constructs a multiple linear regression relationship model between dependent variables and independent variables, establishes an MLR model by using the stepwise regression method, obtains the classified price index affecting rural CPI, and determines its correlation variables. In order to further optimize the correlation coefficient and make the prediction effect of the model more consistent, the GM(1,N) model is further constructed for prediction; Then the data is smoothed, the background value of the original model is optimized based on the analysis of the structural form of the background value of the original model, the background value of GM(1,N) model is reconstructed, and the improved grey GM(1,N) model is proposed. Through operation and model comparative analysis, it is found that the improved model can effectively improve the prediction accuracy and has strong practicability in rural CPI prediction.