ROADSAFE PREDICTOR: ML-DRIVEN ACCIDENT SPOT FORECASTING

Abstract

Due to the exponentially increasing number of vehicles on the road, the number of
accidents occurring on a daily basis is also increasing at an alarming rate. With the
high number of traffic incidents and deaths these days, the ability to forecast the
number of traffic accidents over a given time is important for the transportation
department to make scientific decisions. In this scenario, it will be good to analyze the
occurrence of accidents so that this can be further used to help us in coming up with
techniques to reduce them. Even though uncertainty is a characteristic trait of majority
of the accidents, over a period of time, there is a level of regularity that is perceived
on observing the accidents occurring in a particular area. This regularity can be made
use of in making well informed predictions on accident occurrences in an area and
developing accident prediction models. In this paper, we have studied the inter
relationships between road accidents, condition of a road and the role of
environmental factors in the occurrence of an accident. We have made use of ML
techniques in developing an accident prediction model using Support Vector
Machines and some other ML algorithms road accident datasets for the years few
years available in the internet have been made use for this study.