Reducing Road Accidents in India by Predicting Vehicle Defects and Black Spots

Abstract

Open data is the data that anyone can access, use and share. Many governments have supported the initiative by opening data. Open data supports public oversight of governments and helps reduce corruption by enabling greater transparency. So, government of India has supported the open data by opening several datasets from different departments. These data are available on Open Government Data (OGD) Platform—data.gov.in. We have used these data with the focus of reducing road accidents in India, especially in an area where there is a good scope for public–private cooperation. That is studying road accidents across India happened due to vehicle defects. OGD has categorized many reasons for occurring accident. Some of them include—consumption of alcohol, due to weather condition, junction point, vehicle defects, speed, etc. The main aim of the present study is to build a predictive model for road accident considering previous year’s datasets and predicting the future result. The datasets are collected from the data.gov.in Web site. A datasets consisting of tens of thousand accident records were analyzed and mathematical models were developed by using linear regression. Three factors of accident severity have been examined. The first factor is predicting the result of next year for all states. Then, the second factor is on clustering of states based on high level and low level frequency of accidents. Finally, the third factor is state-wise comparison of accident rate. Inferences are made on some recommendations where private companies join hands with government on keeping the vehicle health data and sharing the information with government and insurance companies, and we have used statistical regression rules and clustering. Estimator will also include error in measure of predicted values by using mean squared error (MSE) and root mean square error (RMSE) methods.