Comparative assessment of radial basis function neural network and multiple linear regression application to trip generation modelling in Akure, Nigeria

Efficacy of using Radial Basis Function Neural Network (RBFNN) and Regression Models (MLR) to estimate trip generation rates in Akure, Nigeria was compared. This sterns from a desire to test more novel modelling techniques besides regression which has hitherto been used in the study area. Data for the study were collected through household questionnaire interview survey in the study area between October 2017 and January 2018. SPSS 22 was used in carrying out data analysis. Correlation analysis showed that Number of household members, (N [subscript HM]), Number of employed household members, (N [subscript EHM]), Number of students in household (N [subscript SH]), Number of Household members with age greater than 12 years, (N [subscript HM12]) and Number of Driver’s license holders in the household, (N [subscript DLH]) were the household variables having significant influence on home based trips generation rates. The models were compared and validated using their R² values and Relative Error (RE). Modelling results showed that RBFNN displayed higher accuracy with R² value of 0.947 and RE of 0.391 as compared to MLR with R² value of 0.589 and RE of 0.875. The study was able to uphold the capability of artificial neural networks to produce better results in travel demand forecasting areas than regression techniques.

Language

  • English

Media Info

Subject/Index Terms

Filing Info

  • Accession Number: 01781975
  • Record Type: Publication
  • Files: TRIS
  • Created Date: Nov 19 2019 2:06PM