Tabular Data with Class Imbalance: Predicting Electric Vehicle Crash Severity with Pretrained Transformers (TabPFN) and Mamba-Based Models

2025-09-14

AI Summary

This study addresses the dual challenges of class imbalance and tabular data modeling in predicting injury severity in electric vehicle collisions. Leveraging 23,301 real-world crash records from Texas (2017–2023), we propose MambaAttention—a novel deep tabular learning framework that synergistically integrates the state-space model Mamba with self-attention to enable dynamic feature reweighting, thereby enhancing detection of rare severe-injury instances. To further mitigate imbalance, we incorporate XGBoost- and Random Forest-based feature selection alongside SMOTEENN hybrid resampling. Experimental results demonstrate that MambaAttention achieves an F1-score of 0.82 on severe-accident classification—outperforming state-of-the-art baselines including TabPFN, MambaNet, and conventional tree-based models. This work advances traffic safety risk prediction by establishing a robust, interpretable, and high-performance deep learning architecture tailored for imbalanced tabular crash data.

Technology Category

• Machine Learning: Feature Construction/Reformulation
• Computer Vision: Multi-modal Vision
• Intelligent Robots: State Estimation

Application Category

• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs

Abstract

This study presents a deep tabular learning framework for predicting crash severity in electric vehicle (EV) collisions using real-world crash data from Texas (2017-2023). After filtering for electric-only vehicles, 23,301 EV-involved crash records were analyzed. Feature importance techniques using XGBoost and Random Forest identified intersection relation, first harmful event, person age, crash speed limit, and day of week as the top predictors, along with advanced safety features like automatic emergency braking. To address class imbalance, Synthetic Minority Over-sampling Technique and Edited Nearest Neighbors (SMOTEENN) resampling was applied. Three state-of-the-art deep tabular models, TabPFN, MambaNet, and MambaAttention, were benchmarked for severity prediction. While TabPFN demonstrated strong generalization, MambaAttention achieved superior performance in classifying severe injury cases due to its attention-based feature reweighting. The findings highlight the potential of deep tabular architectures for improving crash severity prediction and enabling data-driven safety interventions in EV crash contexts.

Problem

• Predicting electric vehicle crash severity with class imbalance
• Benchmarking deep tabular models for severity classification
• Addressing data imbalance in EV crash severity prediction

Innovation

• TabPFN transformer for tabular prediction
• MambaAttention with feature reweighting mechanism
• SMOTEENN resampling for class imbalance