Statistical Analysis of Road Accidents

Abstract

Road accidents are a critical public safety concern, leading to premature deaths, serious injuries, and significant economic losses. Although total prevention may not be feasible, understanding crash patterns through scientific analysis can significantly reduce accident rates. The primary objective of this study is to identify black spots in urban Kurnool by analyzing accident data from the 2-Town Police Station. A research gap was identified in the limited use of collision and condition diagrams in the literature, despite their practical utility in understanding accident causes and geometrical constraints. The study uses a structured methodology involving regression analysis, goodness-of-fit tests, and distribution fitting (Poisson and Negative Binomial) to model accident data from 2019 to 2023. Key findings reveal that the negative binomial distribution better fits the accident data due to over-dispersion. High R² (0.9612) and multiple R² (0.980) values confirm strong predictive accuracy. The study concludes that black spot identification and the visualization of accident patterns through collision and condition diagrams can significantly aid policymakers in implementing focused and cost-effective road safety interventions.