Predicting Fatality Rates Of Road Accidents In Bangladesh Using Artificial Neural Network

Abstract

In recent years, the escalation of road accidents has emerged as a significant global concern, ranking as the ninth leading cause of death worldwide. Bangladesh, in particular, grapples with a substantial burden of road accidents, a distressing reality that demands urgent attention. Recognizing the imperative to address this pressing issue comprehensively, this research paper undertakes a meticulous analysis. The aim is to delve deeper into traffic accidents in Bangladesh, employing machine learning methodologies to discern the magnitude of these incidents. Specifically, an Artificial Neural Network (ANN) is developed, utilizing the ‘relu’ activation function and data spanning from 1998 to 2013 encompassing accidents and casualties. This ANN is trained to forecast the severity of accidents from 2014 to 2024. The model is trained using 70 percent of the data as the training set, with the remaining 30 percent reserved for testing purposes. Through rigorous experimentation, the model is finetuned by varying the number of neurons in a single hidden layer across 1000 epochs, identifying the configuration that yields optimal performance by using the value of RMSE (Root Mean Square Error). Subsequently, the model is trained with this optimal neuron count and applied to predict accident and casualty data from N1 to N9