Classification and Notification of Road Incidents Using Ensemble Approach

Abstract

Normally many vehicle travels on road and number of traffic incidents occur which influences people journeying on that direction. The task of manual classification of these incidents takes more time which may be minimized the use of the system which classifies these incidents using machine learning algorithms. When the incident takes place administrator fills the information of it for dataset generation. This dataset is used for prediction of incident severity. The incidents are categorized for finding out severity using Naïve Bayes and k- Nearest Neighbour classifiers in an effort to discover and visualize frequent patterns in ancient incidents. The classification model accuracy will be progressed using k-fold cross validation. The ensemble approach of machine learning allows to apply different model which yields better results and reduce problem of overfitting by the combination of model. The system has an android software for consumer who's visiting. Even as traveling at the route, person gets notification concerning incidents like accident prone sector, slippery road beforehand and so on. In order that consumer can take suitable choices and saves time throughout touring.