Spatial and factor analysis of vehicle crashes in Mississippi state

Abstract

Traffic crash data from 2010 to 2014 were collected by Mississippi Department of Transportation (MDOT) and extracted for the study. Spatial and factor analysis was conducted including geographic distribution of crashes, descriptive statistics of crash data, and probability analysis of crash factors. Geographic Information System was applied for spatial analysis to show the historical crash data statewide distribution, crash distributions on primary and secondary road segments in the public road system, and crash distribution in MDOT maintenance districts. The results show a similar distribution pattern in the three crash severities in Mississippi as in other states. It also shows that large numbers of the crashes happened on specific locations and there are high crash frequencies on highway segments in metropolitan area and urban area. Based on the spatial analysis results, three comparison scenarios were investigated to estimate the probability of each possible causing factor to the crash severity level. The Type III analysis of variance approach was adopted to assess the significance level of each crash factor, and the multinomial logit model approach with maximum likelihood estimate was applied to conduct the probability analysis and evaluate the significance of each crash factor. The strategies that may potentially decrease the crash frequencies at crash severity levels were discussed based on the probability analysis results.