Using Mining Techniques to Develop a Road Accident Prediction Model

Abstract

A daily tally of accidents is climbing at a frightening pace, paralleling the exponential growth in the volume of automobiles on the street. Given the current high rate of traffic-related events and fatalities, it is crucial for the transportation department to have the capability to envisage the frequency of traffic accidents within a certain timeframe in order to make data-driven choices. In this case, it would be wise to study accident rates in order to utilize that information to develop strategies for lowering them. Although most accidents include some degree of uncertainty, there does seem to be some pattern to the incidents that happen in the same place over time. By utilizing this pattern, ML designs can be constructed for predicting unfortunate incidents and arrive at informed estimations regarding the frequency of mishaps occurring in a specific area. The present study explored the relationships among roadway circumstances, environmental factors, and the number of crashes. A predictive model for accident prediction has been developed by employing the AdaBoost machine learning algorithm. . Various online datasets covering traffic accidents in Bangalore from 2014 to 2017 were used for this research. Public works agencies, contractors, and other sectors of the automotive industry are just a few of the many potential beneficiaries of this study's findings, which may inform more accurate road and vehicle design.