ABSTRACT

India generally represents pretty much one per cent of the worldwide vehicle populace. Be that as it may, it represented around six per cent of the absolute worldwide street mishaps. A sum of 4,37,396 street mishap cases was accounted for during 2019.

We intend to fabricate an expectation model, which will be prepared on the state-wise dataset, for various classifications like age gathering of the driver, climate conditions, street conditions and vehicular imperfections, the season of movement.

In light of this prepared model, when a person enters their course of movement, age of the driver, state of the vehicle and season of movement, the model will anticipate the likelihood of mishap and what could be the reasonable justification of mishaps. We are gathering distinctive state-wise datasets and will group more than 4 diverse datasets to accomplish an adaptable and huge dataset with different contributing components. We intend to lead exploratory information examination after the information cleaning stage. This will assist us with acquiring bits of knowledge about major contributing components for each state.

Then based on the correlation of different parameters a prediction model will be chosen. Using a Client Interface, information will be collected. The prediction model will run on a server and output will be shared on the Client Interface.

INTRODUCTION

India tops the world in road crash deaths and injuries. It has 1 per cent of the world's vehicles but accounts for 11 per cent of all road crash deaths, witnessing 53 road crashes every hour; killing 1 person every 4 minutes. 1214 road crashes occur every day in India. Two-wheelers account for 25% of total road crash deaths. Twenty children under the age of 14 die every day due to road crashes in the country. The average road quality index for India is rated 3.5/7.

This indicates that road accidents are inevitable because some factors are beyond the capacity of human control. But precautionary measures can be taken based on some prediction model. Based on the prediction, route of travel, time of travel, the driver can be changed so that history doesn't repeat itself and accidents are avoided.

OBJECTIVES

With this project we aim to study the causes of road accidents in the country based on past data, we perform an in-depth study on each state, with consideration of different causes such as Vehicular defects, Road defects, weather conditions etc.

Finally, we build a prediction model in R based on the Random Forest prediction algorithm. We host the web-app so it is free and open to use.

Literature Survey

FDA

Finally after the description analysis for each state based on various factors, we build a R-Shiny web app to predict accidents.

We read the data from user in the form of various factors that might cause the accident such as road condition, vehicular condition, state of travel and we read the Weather of that place real-time using Open weather-map API.

We use these predictor variables to predict the severity of road accidents on a scale of 1-7 ranging from very low risk to very high risk.

The link for the web app with prediction: https://ritikagrawal06.shinyapps.io/RoadAccidentVisualizationShinyApp/

Page updated

Google Sites

Report abuse