Zero-inflated regression models for measuring accident

• Main Authors: Hartatik, Nurani, Prasetijo, Joewono, Prasetyo, Yudi Dwi, Nistrina, Khilda, Muslihati, Atqiya
• Format: Conference or Workshop Item
• Language: English
• Published: 2024
• Subjects: T Technology (General)
• Online Access: http://eprints.uthm.edu.my/11122/1/P16673_f522b39b4802ac55a9aadb8b64f06519%206.pdf; http://eprints.uthm.edu.my/11122/; https://doi.org/10.1063/5.0180308

Worldwide, hundreds of thousands of deaths and thousands more are injured every year in traffic accidents around the world. This is owing to an increase in road traffic throughout time, as well as a wide range of traffic compositions. Nowadays, road accidents have become a major concern, and analyzing accidents data has become an important concern for analysts. Therefore, analysis of accidents data requires a lot of attention because accident data is very complex. The road accidents process results in various frequency calculations, for example, deaths and injured number, and/or involved cars in the accidents. However, the probability distribution governing the occurrence of this count may be different. In addition to the problem of excess zeros, lack of data is a common occurrence in results of traffic accidents. Thus, this study discusses the use of the zero-inflated model in analyzing traffic accidents and the variables used by the researchers by reviewing the literature related to the use of zero-inflated models in accident cases. To find a better zero-inflated model that can be used to calculate accident data and to identify the variables that are commonly used to calculate traffic accidents. The result showed that the models that are more widely used by researchers to calculate traffic accidents are commonly known as (ZINB) the zero-inflated negative binomial model and (ZIP) the zero-inflated Poisson model. Both model types have been used since they are approaches for resolving the problem of overdispersion