Analysıs Of Road Traffıc Accıdents Usıng Data Mınıng Technıques

Abstract

Roadway traffic accident (RTA) is one of the major issue faced by almost all of the citizens in the world. Road fatal deaths are constantly increasing while the road safety is seriously lacking and becomes a major concern of transportation agencies. A retrospective observational study was made on data from 2010 to 2017 with an objective to find the injury status based on the severity of the casualties. In this work, we apply statistical study  and data analysis techniques  to address this accident issues. Analysis shows that the distribution of road accidental injuries and deaths varies according to gender, age, surface conditions, weather conditions and lighting conditions. Apriori algorithm is applied to find the interesting association rules. Non linear classification model was built by Naive Bayes classifier, K Nearest Neighbour, Random Forest, Decision Tree, Stochastic Gradient Descendant, XG Boost, Support Vector Machine and Logistic Regression. Hence safe driving recommendations were given depending on the association rules, categorization models and statistics. Achieved result from Support Vector Machine illustrated accuracy at a better level and found variety of  hidden  situations that would be helpful to reduce accident  ratio in future. Thus the preventive measures are to be taken by every private and public sectors, government and non-governmental organizations for an accident less society.