| dc.description.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 | en_US |
