多元线性回归模型应用实证分析

Empirical Analysis of Multiple Linear Regression Models

以福建省物流需求为研究对象,选取福建省1996-2015年货物吞吐量、全社会固定资产投资总额、农林牧渔总产值、进出口商品总额、居民消费水平、地区生产总值等统计数据,基于多元线性回归思路,诊断多重共线性。采用逐步回归法,以SPSS软件为工具,建立货物吞吐量与全社会固定资产投资总额以及进出口商品总额的二元线性回归模型。通过统计分析检验模型的有效性,运用货物吞吐量观测值与预测值的残差及相对误差检验模型的实践性。

This paper takes the logistics demand of Fujian Province as the object of study,and chooses cargo throughput,total social fixed asset investment, total output values of agriculture, forestry, pasturage and fishery, total import and export commodities, household consumption level and regional GDP of Fujian province as the data indicator to diagnosing multiple commonalities based on the multiple linear regression. Using stepwise regression method and SPSS software to establish binary linear regression model of cargo throughput with the total value of total social fixed asset investment and total import and export commodities. The validity of the model is verified by statistical analysis. The residual and relative error of the cargo through put observation value and the predicted value is applied to test the practicality of the model.