CausalTrans: Fusing Causal Inference and Large Language Models for Event-Driven Traffic Forecasting.

Developing a novel deep learning framework to integrate textual event data (e.g., accidents, construction) for causal-aware traffic flow prediction, aiming to improve model robustness and interpretability.

Traffic Prediction via Signal Decomposition and Attention-based BiLSTM

Developed a dual-stream BiLSTM model incorporating moving average, Fourier transform, and multi-head attention mechanisms to capture trend and seasonal components in traffic flow data by integrating signal decomposition with deep learning frameworks.

Behavioral Changes from EV Adoption: A Quasi-Experimental Study

Designed and executed a quasi-experimental study using mobile signaling data to assess causal effects of electric vehicle (EV) adoption on travel behavior, comparing EV adopters with internal combustion engine (ICE) vehicle users as controls.

Synthesizing the Impact of Travel Behavior Interventions: A Meta-Analysis

Conducted a systematic review and meta-analysis of 24 global studies (38 effect sizes) to quantify the effectiveness of travel behavior interventions, addressing fragmented knowledge and identifying key factors for public transport policy success.

LLM Analysis of Policy Impact on Cross-Border Spatial Sentiment

Bridged gaps in spatial sentiment quantification and causal inference by developing a pipeline integrating LLM-based sentiment analysis with interrupted time series (ITS) modeling to evaluate regional integration policy impacts.

Joint Design of SAV-Transit Systems for Tourist Cities

Explored integrated design of public transit and shared autonomous vehicle (SAV) systems for tourist cities, accounting for private tour trips, demonstrating up to 25% cost reduction in low-demand scenarios.

Travel Route Choice Preference Analysis Based on Phone Signaling Big Data

Employed machine learning models (SVM, XGBoost, Random Forest) to analyze factors influencing route choices, with emphasis on congestion impacts, enabling accurate monitoring of congestion states and preference features from mobile signaling data.

Optimized Path Planning and Traffic Flow Distribution Model

Developed a model to mitigate navigation-induced traffic aggregation and user dissatisfaction, balancing user equilibrium with system optimization for enhanced network efficiency.