Comparative Analysis on the Prediction of Road Accident Severity Using Machine Learning Algorithms

Abstract

AbstractThe present expansion ratio of the human population is a major concern, and it is ever growing. In peak hours, traffic congestion is high and may lead to road accidents. Most of the fatal road accidents are due to unruly driving, disturbing reflections from fellow vehicles and other factors. Such incidents cause loss of human lives and properties. This paper studies multilevel models such as prediction and classification algorithms which are analyzing the severity of the accidents. Further, this will help to minimize casualties and follow established road safety measures. Prediction algorithm is used for predicting the occurrence of road accidents, and classification algorithm is used for categorizing the severity of road accidents into fatal, severe and mild injury. The researchers have used different machine learning (ML) algorithms that would analyze the severity of road accidents and other vital risk factors. ML models are also developed with the help of Internet of things (IoT) for prediction and classification. In this work, comparative study is done on the basis of performance metrics as accuracy on different ML algorithms. It is found that the random forest (RF) algorithm performs better than other algorithms. These algorithms cater high-quality results for the accident database. This study may assist the researcher, Public Works Department (PWD), and be useful to build government policies for identifying vital risk factors and minimizing road accidents.KeywordsPredictionClassificationSeverityMachine learningInternet of things