摘要
以印度尼西亚首都雅加达都市圈居民个人出行调查数据为例,研究EM数据修补方法对数据以及MNL模型的影响.首先,以原始数据为基础,通过人为删除和EM修补分别获得缺失数据和修补数据.其次,通过Z检验,验证EM修补后的数据更贴近原始数据特征.最后,以三组数据分别建立三组MNL模型,通过Z检验等对比分析,表明EM数据修补方法能很好地修正数据缺失对构建模型造成的偏差,为交通政策的制定提供良好的数据基础.
This paper explores the influences of EM imputation on data and MNL models based on personal trip data collected in Jabodetabek metropolitan area,Indonesia. First,missing dataset and imputed dataset are obtained by manually deleting the cases of complete original data and EM imputation,respectively. Secondly,dataset by EM imputation is verified to be more close to the original dataset by statistics Z test. Finally,the analysis such as Z test is conducted to compare three MNL models built on original dataset,missing dataset and imputed dataset. The result reveals that EM imputation can effectively correct the bias caused by missing data in modeling building,which could offer a good data base for policy making.
出处
《大连交通大学学报》
CAS
2017年第3期7-11,共5页
Journal of Dalian Jiaotong University
基金
中央高校基本科研业务费专项资金资助项目(3132016213)
