TabReason: A Reinforcement Learning-Enhanced Reasoning LLM for Explainable Tabular Data Prediction

TL;DR

TabReason integrates reinforcement learning with reasoning-based large language models to enhance the accuracy and interpretability of tabular data predictions, addressing the limitations of existing models in explainability.

Contribution

It introduces a novel reinforcement learning framework that trains LLMs to produce more accurate and explainable predictions on tabular data.

Findings

Improved prediction accuracy on financial datasets.

Enhanced interpretability through human-understandable explanations.

Outperforms baseline LLMs in benchmark tests.

Abstract

Predictive modeling on tabular data is the cornerstone of many real-world applications. Although gradient boosting machines and some recent deep models achieve strong performance on tabular data, they often lack interpretability. On the other hand, large language models (LLMs) have demonstrated powerful capabilities to generate human-like reasoning and explanations, but remain under-performed for tabular data prediction. In this paper, we propose a new approach that leverages reasoning-based LLMs, trained using reinforcement learning, to perform more accurate and explainable predictions on tabular data. Our method introduces custom reward functions that guide the model not only toward better prediction accuracy but also toward human-understandable reasons for its predictions. The proposed method is evaluated on financial benchmark datasets and compared against established LLMs.