大数据背景下网络调查样本的随机森林倾向得分模型推断研究

随着大数据与互联网技术的迅猛发展,网络调查的应用越来越广泛。本文提出网络调查样本的随机森林倾向得分模型推断方法,通过构建若干棵分类决策树组成随机森林,对网络调查样本单元的倾向得分进行估计,从而实现对总体的推断。模拟分析和实证研究结果表明:基于随机森林倾向得分模型的总体均值估计的相对偏差、方差与均方误差均比基于Logistic倾向得分模型的总体均值估计的相对偏差、方差与均方误差小,提出的方法估计效果更好。

Research on Random Forest Propensity Score Modeling Inference of Web Survey Samples Under the Background of Big Data

With the rapid development of big data and internet technology, web surveys have been widely used. This paper proposes an inference approach that adopts random forest propensity score modeling, in which propensity scores of web survey sample units are estimated through a random forest that consists of several classification decision trees for inference of the population. The results from simulation analysis and empirical research show that the relative biases, variances and mean square errors of the population mean estimator based on the random forest propensity score model are all smaller than those based on the logistic propensity score model, indicating that the proposed method has better performance.