Generalized Linear Model Approach for Road Accident Patterns in Kerala, India

Abstract

India's development over the past few years and an individual's desire to own vehicles has led to increased motorization across different states in India, with Kerala being one among them. In today's competitive world, everyone is in hurry and their impatience has resulted in careless attitudes towards driving, as well as towards vehicle maintenance, which in turn has risen the number of fatalities due to road accidents in Kerala. In this study we have made an effort to determine regions that are at greater risk of fatality and the vehicles that most prominently contribute to such risk using the Generalized Linear Model (GLM) approach. On comparison, Negative Binomial Regression Model (NBRM) showed up to be more appropriate over Poisson Regression Model (PRM). With Wayanadu as the baseline district, Idukki reported the lowest (1.303) incident rate. Among Ernakulam, Kozhikkode and Trivandrum, urban areas were exposed to lower fatality risk as compared to rural areas. Further, 'KSRTC bus', 'Cars & Two-wheelers' were found to cause more fatal accidents with 0.1% and 0.001% increase in deaths for an increment of one case due to each of them respectively. The prediction model adopted, thus yielded convincing results, which can be utilized for improving not only the human but economic and environmental conditions in Kerala.