基于广义估计方程的时间相关性事故频次建模（英文）

发布时间：2018-10-30 18:19

【摘要】：研究目的:采用广义估计方程模型对存在时间相关性的事故频次数据进行建模,并与传统广义线性模型的估计效果进行对比。创新要点:通过广义估计方程来考虑事故频次建模中数据的时间相关性,从而提高参数估计准确度以及模型预测精度。研究方法:基于4年高速公路交通事故频次数据,建立考虑时间相关性的广义估计方程以及传统的广义线性模型,并采用统计指标对模型效果进行对比。重要结论:1.事故频次数据样本量对预测精度影响很大;2.广义估计方程能够有效考虑事故频次数据中存在的时间相关性;3.广义估计方程的参数估计比传统广义线性模型更准确,且精度更高。
[Abstract]:Objective: to model the time-dependent accident frequency data by using the generalized estimation equation model and compare the estimation results with the traditional generalized linear model. Innovation points: the time correlation of the data in accident frequency modeling is considered through the generalized estimation equation, so as to improve the accuracy of parameter estimation and the prediction accuracy of the model. Methods: based on the frequency data of highway traffic accidents for 4 years, the generalized estimation equation and the traditional generalized linear model were established, and the statistical indexes were used to compare the effects of the model. Important conclusions: 1. The sample size of accident frequency data has a great influence on the prediction accuracy; 2. The generalized estimation equation can effectively consider the time correlation in accident frequency data. The parameter estimation of the generalized estimation equation is more accurate and accurate than the traditional generalized linear model.
【作者单位】： Jiangsu
【基金】：Project supported by the National Key Basic Research Program(973)of China(No.2012CB725400) the National High-Tech R&D Program(863)of China(No.2012AA112304) the National Natural Science Foundation of China(No.51338003)
【分类号】：U491.31

【相似文献】