作者: Lu Ma , Xuedong Yan , Chong Wei , Jiangfeng Wang
DOI: 10.1016/J.AMAR.2016.07.001
关键词: Poison control 、 Log-normal distribution 、 Econometric model 、 Order (exchange) 、 Engineering 、 Tobit model 、 Regression analysis 、 Econometrics 、 Regression 、 Loss and damage 、 Simulation
摘要: Abstract The understanding of the distributional characteristics equivalent property damage only (EPDO) crash rate is limited in existing literature. Models without a proper distribution EPDO could result biased estimations and misinterpretations factors. importance prediction accuracy modeling performance for should be acknowledged since they directly affect allocation public funds to safety management road networks. general objective this study investigate accordingly develop econometric models connecting explanatory variables. A hurdle framework was proposed order accommodate zero-positive mixed domain rate. For positive part rate, three representative distributions (lognormal, gamma normal) were tested then compared against Tobit model random-parameters model. empirical results illustrate lognormal model's superior comparison other four models, more importantly that conclusion also holds several different definitions under combinations (PDO) equivalency