Citation

Irshaid M, Abu-Eisheh S. Ain Shams Eng. J. 2023; 14(11): e102523.

Copyright

(Copyright © 2023, Ain Shams University, Publisher Elsevier Publishing)

DOI

10.1016/j.asej.2023.102523

PMID

unavailable

Abstract

This study aims to investigate the feasibility of using the Adaptive Neuro-Fuzzy Inference System (ANFIS) and Multiple Linear Regression (MLR) for modelling home-based trip generation in Salfit City, Palestine. The research compares the performance of these two methods and provides insights into their efficiency for different trip purposes. The methodology involves developing separate trip generation models for various purposes, including total daily household trips generated (HBALL), home-based work (HBW), home-based education (HBE), and other home-based trips (HBO). The underlying principles and implementation details of the ANFIS architecture are also discussed.

OBJECTIVE evaluation metrics like Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), and R-squared are used to assess the effectiveness of each model. The results indicate that ANFIS performs well for modeling HBALL and HBO trips, which exhibit more complex behavior with wider data ranges and higher average daily trip counts. Compared to the MLR approach, ANFIS shows improved accuracy and closer predictions. For instance, when modelling HBALL trips, ANFIS achieved an RMSE of 1.4880, while MLR resulted in 1.7112, representing a reduction of 13.04%. However, for HBW and HBE trip purposes, where behavior is less complicated, MLR appears to be sufficient. The R-squared values obtained with MLR are sufficiently high to capture most of the trip variations, and the results between the two approaches are closely comparable. The R-squared value for MLR in HBE trips was almost identical to that of ANFIS, which was 96.65%. Similarly, for HBW trips, MLR achieved an R-squared value of 90.36%, while ANFIS performed slightly better with an R-squared value of 92.74%. In conclusion, ANFIS shows promise for modelling systems with complex behavior, while MLR remains a suitable option for less complicated scenarios. The study emphasizes the importance of exploring different modelling techniques in transportation research to identify the most appropriate approach for specific cases.

Language: en

Keywords

Adaptive neuro-fuzzy inference system; Home-based trip generation; Multiple linear regression; Travel demand modelling