PREDICTIVE CRASHES MODELS ON SELECTED RURAL ROAD NETWORKS: A CASE STUDY OF AKURE-OWO AND ONDO ROAD, AKURE ONDO STATE

Abstract

i. Road transportation, a major means of conveying goods and services in developing countries like Nigeria, has led to rapid increase in motorization without provision for adequate road traffic crashes and injuries control mechanisms. The main research goal of this study is to develop prediction models and to evaluate the influence of various causative factors of crashes on two-lane rural roads. Two rural roads namely Akure-Owo (AO1) and Akure-Ondo (AO2) road critical to social economic development in Ondo State, South-West Nigeria were carefully selected for detailed study. A reconnaissance survey was conducted to identify the black spots followed by detailed study to obtain traffic and geometric data at the identified locations. Alignment indices of black spots were obtained from the road profile plotted using Geographical information software packages (Goggle Earth, Global Mapper) and Civil 3D. Crash data from the year 2010 to 2015 was collected from Federal Road Safety Corps (FRSC) Ondo State Sector Command. The descriptive and linear regression analysis of variables factors (Driver factors, Vehicle factors, Roadway factors and Environmental factors) contributing to road crashes was carried out using Microsoft Excel. The analysis of the crash data shows 11% decrease in crash and 22% decrease in fatality trend from 2010 to 2015 along AO1 while, there is 20 % and 30% decrease in crash and fatality trend in AO2. The study revealed that 487 crashes occurred between 2010 to 2015 involving 3,533 people which lead to 329 deaths on AO1. On AO2 road, 97 crashes involving 526 occurred leading to 56 deaths. The descriptive analysis shows that driver factors (X1), mechanical/vehicle factors (X2), road factors (X3) and environmental factors (X4) account for 80%, 16%, 4% and 1% respectively on AO1; on the other hand, on AO2, 87%, 10%, 2% and 1% accounted for the aforementioned causative factors. 99.4% and 99.7% was obtained for the coefficient of correlation “R” while, 98.6% and 98.44% was obtained for the coefficient of determination for AO1 and AO2 respectively. The study further revealed that driver’s, vehicle and roadway factors were statistically significant at 5% level of significant; however, driver and vehicle factors accounted for 90% of the road crashes along AO1 and AO2. Furthermore, the coefficient of determination (R2) value of 99.41% was obtained from the model formulated for the selected roads; this indicates that the causative factors were highly significant to the model. On the other hand, the parameters obtained from black spots were used to formulate negative binomial regression models that relate the crashes to traffic parameters and alignment indices using Statistical and Data software (STATA version 12). The model AO1 revealed that right shoulder and degree of curvature were significant to road crashes with (𝑝<0.05). Right shoulder had a positive sign effect ( 𝛽=0.4364768), signifying that a unit increase in right shoulder width will cause an increase in crash frequency if other variables are kept constant. Furthermore, the model revealed that degree of curvature was also significant to road crashes along AO1 with a negative effect (𝛽=−0.2632294), this indicate that a unit increase in degree of curvature will cause a unit decrease in crash rate. The AO2 model showed that traffic and geometric indices were not significant with crashes along Ondo road with (𝑝>0.05). The models formulated were validated and in order to promote safety along these roads, the degree of curvature of black spots should be widening, existing shoulder should be re-graded and paved, Drivers and other road users should be properly trained and evaluated before being certified to drive in all the highways and strict enforcement of road traffic laws on offenders