Expert Knowledge-Guided Travel Demand Estimation: Neuro-Fuzzy Approach

Abstract

Travel demand estimation is a complicated process because it is highly influenced by human perceptions and behaviors embedded in travelers’ decision making. In addition, some variables in travel demand estimation models have inexact values, especially the ones that are required for future estimation of travel demand. This article offers and applies an expert-guided algorithm to incorporate expert knowledge into Adaptive Network-Based Fuzzy Inference System (ANFIS) for handling travel demand estimation uncertainties. ANFIS is an appropriate hybrid intelligent system combining fuzzy logic and neural networks. The ANFIS rule base structure facilitates integration of qualitative knowledge acquired from experts and quantitative information from observational data. An expert-guided algorithm in conjunction with ANFIS, called expert-guided ANFIS (EGANFIS), is presented to compensate data insufficiency caused by uncertainties. The EGANFIS is applied to a real-world problem for estimating trip production, attraction, distribution, and modal split in Shiraz, a large city in Iran. The comparison of results with traditional models and ANFIS shows that EGANFIS increases performance of travel demand estimation models in terms of learning and indicates that the model is accurate enough to provide meaningful information and to enable generalization of the findings. Furthermore, rule base structure of EGANFIS enhances its interpretability in comparison with traditional models.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 13 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.