Modeling road accident fatalities with underdispersion and zero-inflated counts

被引:6
|
作者
Simmachan, Teerawat [1 ,2 ]
Wongsai, Noppachai [2 ]
Wongsai, Sangdao [1 ,2 ]
Lerdsuwansri, Rattana [1 ]
机构
[1] Thammasat Univ, Fac Sci & Technol, Dept Math & Stat, Pathum Thani, Thailand
[2] Thammasat Univ, Res Unit Data Learning, Pathum Thani, Thailand
来源
PLOS ONE | 2022年 / 17卷 / 11期
关键词
TRAFFIC ACCIDENTS; REGRESSION-MODEL; SEVERITY; THAILAND; CRASHES; INJURY; COST;
D O I
10.1371/journal.pone.0269022
中图分类号
O [数理科学和化学]; P [天文学、地球科学]; Q [生物科学]; N [自然科学总论];
学科分类号
07 ; 0710 ; 09 ;
摘要
In 2013, Thailand was ranked second in the world in road accident fatalities (RAFs), with 36.2 per 100,000 people. During the Songkran festival, which takes place during the traditional Thai New Year in April, the number of road traffic accidents (RTAs) and RAFs are markedly higher than on regular days, but few studies have investigated this issue as an effect of festivity. This study investigated the factors that contribute to RAFs using various count regression models. Data on 20,229 accidents in 2015 were collected from the Department of Disaster Prevention and Mitigation in Thailand. The Poisson and Conway-Maxwell-Poisson (CMP) distributions, and their zero-Inflated (ZI) versions were applied to fit the data. The results showed that RAFs in Thailand follow a count distribution with underdispersion and excessive zeros, which is rare. The ZICMP model marginally outperformed the CMP model, suggesting that having many zeros does not necessarily mean that the ZI model is required. The model choice depends on the question of interest, and a separate set of predictors highlights the distinct aspects of the data. Using ZICMP, road, weather, and environmental factors affected the differences in RAFs among all accidents, whereas month distinguished actual non-fatal accidents and crashes with or without deaths. As expected, actual non-fatal accidents were 2.37 times higher in April than in January. Using CMP, these variables were significant predictors of zeros and frequent deaths in each accident. The RAF average was surprisingly higher in other months than in January, except for April, which was unexpectedly lower. Thai authorities have invested considerable effort and resources to improve road safety during festival weeks to no avail. However, our study results indicate that people's risk perceptions and public awareness of RAFs are misleading. Therefore, nationwide road safety should instead be advocated by the authorities to raise society's awareness of everyday personal safety and the safety of others.
引用
收藏
页数:23
相关论文
共 50 条
  • [21] Interpretable zero-inflated neural network models for predicting admission counts
    Jose, Alex
    Macdonald, Angus S.
    Tzougas, George
    Streftaris, George
    ANNALS OF ACTUARIAL SCIENCE, 2024, 18 (03) : 644 - 674
  • [22] Zero-inflated Poisson factor model with application to microbiome read counts
    Xu, Tianchen
    Demmer, Ryan T.
    Li, Gen
    BIOMETRICS, 2021, 77 (01) : 91 - 101
  • [23] A general location model with zero-inflated counts and skew normal outcomes
    Mirkamali, Sayed Jamal
    Ganjali, Mojtaba
    JOURNAL OF APPLIED STATISTICS, 2017, 44 (15) : 2716 - 2728
  • [24] The LZIP: A Bayesian Latent Factor Model for Correlated Zero-Inflated Counts
    Neelon, Brian
    Chung, Dongjun
    BIOMETRICS, 2017, 73 (01) : 185 - 196
  • [25] Modeling count data with marginalized zero-inflated distributions
    Cummings, Tammy H.
    Hardin, James W.
    STATA JOURNAL, 2019, 19 (03): : 499 - 509
  • [26] Random Effects Modeling and the Zero-Inflated Poisson Distribution
    Monod, Anthea
    COMMUNICATIONS IN STATISTICS-THEORY AND METHODS, 2014, 43 (04) : 664 - 680
  • [27] The analysis of zero-inflated count data: Beyond zero-inflated Poisson regression.
    Loeys, Tom
    Moerkerke, Beatrijs
    De Smet, Olivia
    Buysse, Ann
    BRITISH JOURNAL OF MATHEMATICAL & STATISTICAL PSYCHOLOGY, 2012, 65 (01): : 163 - 180
  • [28] Multivariate zero-inflated modeling with latent predictors: Modeling feedback behavior
    Fox, Jean-Paul
    COMPUTATIONAL STATISTICS & DATA ANALYSIS, 2013, 68 : 361 - 374
  • [29] Zero-inflated hierarchical models for faecal egg counts to assess anthelmintic efficacy
    Wang, Craig
    Torgerson, Paul R.
    Hoglund, Johan
    Furrer, Reinhard
    VETERINARY PARASITOLOGY, 2017, 235 : 20 - 28
  • [30] Modelling Zero-inflated Fish Counts in Estuaries - A Comparison of Alternate Statistical Distributions
    Mayer, D.
    Roy, D.
    Robins, J.
    Halliday, I.
    Sellin, M.
    MODSIM 2005: INTERNATIONAL CONGRESS ON MODELLING AND SIMULATION: ADVANCES AND APPLICATIONS FOR MANAGEMENT AND DECISION MAKING: ADVANCES AND APPLICATIONS FOR MANAGEMENT AND DECISION MAKING, 2005, : 2581 - 2587