Comparison of accident prediction model between ANN and MLR models

Ab. Ghani, Ahmad Raqib and Sanik, Mohd Erwan and Mohd Mokhtar, Roselinda Aida (2011) Comparison of accident prediction model between ANN and MLR models. In: International Seminar on the Application of Science and Mathematics 2011, 1 - 3 November 2011, Putra World Trade Centre, Kuala Lumpur.

[img]
Preview
PDF
99Kb

Abstract

Accident rate that occurred in Malaysia has been increased for every year. Govern ment and all concern parties constantly worrying about this matter where serious measures should be taken to prevent this rising accident rate fro m happened. Therefore, a forecasting of accident prediction models have to be developed. In this study, the locations were focus in rural selected area. The locations of the study were selected among the highest accident rates in Federal Route 050 based on accident point weightage analysis. Traffic volu me, speed, number of access point and gaps data were used to develop the models. Data collections have been done through manual observation at high risk area. The parameters were then used to develop an accident predictio n model by using the Artificial Neural Network (ANN) and Multiple Linear Regression (MLR) models. Both models were then utilized and MLR model was identified to give the better result in term of reducing the number of accidents compared to ANN. Therefore, MLR model was suggested to be used by the concern parties in order to predict the accident as to reduce the accidents more effectively and further to achieve the national set reduction target.

Item Type:Conference or Workshop Item (Paper)
Uncontrolled Keywords:accident prediction model; artificial neural network; multiple linear regression
Subjects:H Social Sciences > HE Transportation and Communications
Divisions:Faculty of Civil and Environmental Engineering > Department of Material and Structure Engineering
ID Code:2372
Deposited By:M.Iqbal Zainal A
Deposited On:05 Mar 2012 10:08
Last Modified:05 Mar 2012 10:08

Repository Staff Only: item control page