论文部分内容阅读
为了提高基本路段事故预测模型(SPF)的预测精度,收集了640个基本路段设计资料及事故资料,应用3个负二项回归模型(NB)和3个广义负二项(GNB)回归模型对收集的数据进行拟合,并分析了解释变量的交互影响.研究表明在上述6个模型中,其中考虑了年平均日交通量和路段长度交互影响的2个模型(一个为NB,另一个为GNB),其预测结果更为合理.进一步综合对比表明考虑交互影响时,NB模型和GNB模型的适用性几乎相同,而GNB略佳.
In order to improve the prediction accuracy of the SPF prediction model, 640 basic section design data and accident data were collected. Three negative binomial regression models (NBs) and three generalized negative binomial (GNB) regression models were used The data collected were fitted and the interaction effects of explanatory variables were analyzed.The results show that in the above six models, two models (one is NB and the other is GNB), the prediction result is more reasonable.A further comprehensive comparison shows that the applicability of the NB model and the GNB model is almost the same while the GNB is slightly better considering the interaction effect.