Towards Locally Deployable Fine-Tuned Causal Large Language Models for Mode Choice Behaviour

2025-07-28

AI Summary

Existing travel mode choice prediction models struggle to simultaneously ensure interpretability and support privacy-preserving local deployment. Method: We propose LiTransMC—the first causal large language model (LLM) fine-tuning framework tailored for this task—incorporating fine-grained parameter-efficient tuning and a loss masking mechanism to achieve high predictive accuracy and near-perfect probability calibration; we further introduce the Explanation Strength Index (ESI) to quantify alignment between model reasoning and behavioral theory. Results: On multiple public multimodal datasets, LiTransMC achieves a weighted F1-score of 0.6845 and a Jensen–Shannon divergence of only 0.000245, significantly outperforming both untuned local LLMs, GPT-4o, and conventional discrete choice models. The framework enables privacy-aware, low-cost, and reproducible transportation behavior analysis, advancing the practical deployment of interpretable, causal AI in evidence-based transport policy design.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Planning, Routing, and Scheduling: Planning with Language Models
• Application Domains: Transportation

Application Category

• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems

Abstract

This study investigates the adoption of open-access, locally deployable causal large language models (LLMs) for travel mode choice prediction and introduces LiTransMC, the first fine-tuned causal LLM developed for this task. We systematically benchmark eleven LLMs (1-12B parameters) across three stated and revealed preference datasets, testing 396 configurations and generating over 79,000 synthetic commuter predictions. Beyond predictive accuracy, we evaluate models generated reasoning using BERTopic for topic modelling and a novel Explanation Strength Index, providing the first structured analysis of how LLMs articulate decision factors in alignment with behavioural theory. LiTransMC, fine-tuned using parameter efficient and loss masking strategy, achieved a weighted F1 score of 0.6845 and a Jensen-Shannon Divergence of 0.000245, surpassing both untuned local models and larger proprietary systems, including GPT-4o with advanced persona inference and embedding-based loading, while also outperforming classical mode choice methods such as discrete choice models and machine learning classifiers for the same dataset. This dual improvement, i.e., high instant-level accuracy and near-perfect distributional calibration, demonstrates the feasibility of creating specialist, locally deployable LLMs that integrate prediction and interpretability. Through combining structured behavioural prediction with natural language reasoning, this work unlocks the potential for conversational, multi-task transport models capable of supporting agent-based simulations, policy testing, and behavioural insight generation. These findings establish a pathway for transforming general purpose LLMs into specialized, explainable tools for transportation research and policy formulation, while maintaining privacy, reducing cost, and broadening access through local deployment.

Problem

• Develop locally deployable LLMs for travel mode prediction
• Evaluate LLM reasoning alignment with behavioral theory
• Enhance predictive accuracy and interpretability in transportation models

Innovation

• Fine-tuned causal LLMs for mode choice
• Parameter efficient loss masking strategy
• BERTopic and Explanation Strength Index