Multi-level Diagnosis and Evaluation for Robust Tabular Feature Engineering with Large Language Models

Yebin Lim, Susik Yoon
9/20/2025
cs.LGcs.AI

No official code found. Best implementation: DohaLim/Robustness-eval (1 star, last commit 2025-12-22).

Abstract

Recent advancements in large language models (LLMs) have shown promise in feature engineering for tabular data, but concerns about their reliability persist, especially due to variability in generated outputs. We introduce a multi-level diagnosis and evaluation framework to assess the robustness of LLMs in feature engineering across diverse domains, focusing on the three main factors: key variables, relationships, and decision boundary values for predicting target classes. We demonstrate that the robustness of LLMs varies significantly over different datasets, and that high-quality LLM-generated features can improve few-shot prediction performance by up to 10.52%. This work opens a new direction for assessing and enhancing the reliability of LLM-driven feature engineering in various domains.

View on arXivView PDF

Code Implementations(1)

DohaLim/Robustness-evalCommunity75%

Multi-level Diagnosis and Evaluation for Robust Tabular Feature Engineering with Large Language Models

11Shell, PythonSep 19, 202510 months ago

Similar Papers

Large Language Models Engineer Too Many Simple Features For Tabular Data

87% match1 repo62024

Human-LLM Collaborative Feature Engineering for Tabular Data

86% match2026

Robustness in Large Language Models: A Survey of Mitigation Strategies and Evaluation Metrics

86% match2025

LLM-FE: Automated Feature Engineering for Tabular Data with LLMs as Evolutionary Optimizers

86% match1 repo812025

ELF-Gym: Evaluating Large Language Models Generated Features for Tabular Prediction

85% match1 repo62024

A Note on Statistically Accurate Tabular Data Generation Using Large Language Models

85% match1 repo2025

Cite this paper

@article{lim2025multilevel,
  title  = {Multi-level Diagnosis and Evaluation for Robust Tabular Feature Engineering with Large Language Models},
  author = {Yebin Lim and Susik Yoon},
  year   = {2025},
  eprint = {2509.25207},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2509.25207v1}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2509.25207.svg)](https://codeofpaper.com/paper/2509.25207)

Discussion