Transport Choice Modeling for the Evaluation of New Transport Policies

Pijoan, Ander; Kamara-Esteban, Oihane; Alonso-Vicario, Ainhoa; Borges, Cruz E. · 2018 · Crossref

DOI: 10.3390/su10041230

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This study addresses the challenge of evaluating sustainable transport policies by improving transport choice modeling, a critical component of traffic simulation that determines how travelers select modes of transport. While traditional simulation tools often rely on discrete choice regression models (such as logit models), these methods require extensive recalibration when new transport modes or policies are introduced. The authors aim to identify more flexible and accurate data science methods for modeling human decision-making in transport choices, thereby facilitating the assessment of policy impacts on greenhouse gas emissions and traffic behavior. The researchers compared eight modeling techniques: k-nearest neighbors (KNN), multinomial logit, support vector machines (SVM), neural networks (NN), naïve Bayes, fuzzy logic, expert knowledge-based fuzzy rules, and random search. The models were trained using commute data from the Biscay province in Spain, derived from census data covering approximately 1.1 million citizens across 11 municipalities. The input features for the models included itinerary-dependent variables: duration, price, length, and environmental impact (CO2 emissions). Missing values in the dataset, attributed to routing algorithm timeouts, were replaced with the maximum observed value. To validate the models' generalizability, the authors tested them on a separate dataset from the Silesian Voivodeship in Poland, which included survey data from 4.7 million citizens across 19 municipalities. The results indicate that the best-performing models correctly forecasted more than 51% of the recorded trips in the training set. Crucially, when validated against the Silesian dataset, all models maintained their forecasting ability, demonstrating robustness across different geographical and demographic contexts. The study highlights that while traditional logit models are widespread, alternative methods like fuzzy logic and neural networks offer viable alternatives for capturing complex traveler preferences. The fuzzy logic approach, in particular, was noted for its interpretability and ability to emulate individual preferences through rule-based systems, though it required optimization via co-evolutionary algorithms to match the performance of black-box models. The significance of this work lies in providing a comparative framework for selecting transport choice models that are both accurate and adaptable. By validating models on distinct datasets, the authors demonstrate that these methods can reliably predict transport mode choices in diverse settings. This capability is essential for policymakers and urban planners who need to simulate the effects of new transport infrastructures, incentives, or restrictions without the prohibitive cost of recalibrating entire simulation models. The findings support the integration of advanced data science techniques into transport planning to better evaluate sustainable mobility policies.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• exposure measurement

Information type

What kind of knowledge this paper contributes, grouped by family — independent of topic (what it is about) and method (how it was studied).

• Theoretical Contribution: computational model