Modeling severity of road traffic accident in Nigeria using artificial neural network

• Main Authors: Umar, Ibrahim Khalil, Gokcekus, Huseyin
• Format: Article
• Language: English
• Published: Penerbit Universiti Kebangsaan Malaysia 2019
• Online Access: http://journalarticle.ukm.my/14815/1/06.pdf; http://journalarticle.ukm.my/14815/; http://www.ukm.my/jkukm/volume-312-2019/
• Institution: Universiti Kebangsaan Malaysia

In this study, an Artificial Neural Network (ANN) was used to model injury and fatality index in Nigeria with the aim to determine the effects of the number of GSM subscription (NGS) on the injury and fatality index in the country. Fifty-seven-year data from 1960-2016 comprising of Gross Domestic Product (GDP) per capita, population, NGS, the total number of traffic accidents, number of fatality and injury per year were used for developing the model. The result of the ANN implies that adding the NGS to the model has increased the model performance in both training and testing with a determination coefficient increasing by 18.7% and 2.5% in testing for fatality and injury index respectively. Comparing the performance of the ANN models and regression analysis shows the superiority of the ANN technique over the regression analysis for both injury and fatality index models. The goodness of fit of the model was further checked using t-test at 5% level of significance and the result proved the ANN approach as a powerful tool for modeling the severity of road traffic accident. Strict enforcement against the use of phone while driving will help reduce the accident severity caused as a result of phone usage while on wheels.