Spatio-temporal analysis of road traffic accidents in Tunisia

Abstract

Focusing on hotspot mapping and comparing different mapping methods can help improve their practical value in the field by better predicting crash patterns. The use of a geographic information system (GIS) could prevent further damage. The aim of this research is to recommend the best GIS technique for accident investigation in different scenarios. In this review article, we introduce and discuss two basic GIS methods for simulating road accidents and offer some recommendations for effective road safety accident analysis tools. Based on the work examined, current issues and future research directions are determined. The purpose of this article is to determine the location and duration of road sections with high accident rates (black zones). A comparison is made between two approaches: one employs the average nearest neighbor to identify highway segments where motorized collisions are clustered, scattered, or random, the other employs kernel density estimation to identify black spots. Our study differs from others in that it examines the identification of potential hotspots and improves the ability to probe a specific lane through identification “hazardous probable lengths,” which aims to predict future traffic crashes. The detected and likely to be identified hotspots have different spatial and temporal characteristics. Different regional and temporal characteristics are present at the identified hotspots. It is clear that there are some geographic clusters of accidents. Most of the identified hotspots in the Northwest and Center-West regions are located along rural highways. In the Central-East region, both hot zones are broadly distributed to the Northeast and Southwest, particularly in NH1 and NH2, where much urban activity occurs. Spatial autocorrelation indices per region address the variability within the areas and provide us with important insights that can feed into Tunisian safety regulations.