Large Language Models for Missing Data Imputation: Understanding Behavior, Hallucination Effects, and Control Mechanisms

2026-03-20

AI Summary

This study addresses critical gaps in existing missing data imputation methods—particularly their limited scalability, lack of cross-model comparability, and insufficient systematic evaluation across diverse missingness mechanisms (MCAR, MAR, MNAR). For the first time, it conducts a large-scale comparison of five large language models (e.g., Gemini 3.0 Flash, Claude 4.5 Sonnet) against six traditional approaches (e.g., MICE), evaluating their performance via zero-shot prompting across 29 real-world and synthetic datasets under varying missing rates and mechanisms. The findings reveal that large language models leverage semantic priors rather than statistical reconstruction, significantly outperforming conventional methods on real data but underperforming on synthetic data. Despite achieving higher imputation quality, these models incur substantially greater computational and economic costs.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: (Large) Language Models
• Computer Vision: Large Vision Models

Application Category

• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Search and Retrieval-Augmented AI: Large language models for search
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

Data imputation is a cornerstone technique for handling missing values in real-world datasets, which are often plagued by missingness. Despite recent progress, prior studies on Large Language Models-based imputation remain limited by scalability challenges, restricted cross-model comparisons, and evaluations conducted on small or domain-specific datasets. Furthermore, heterogeneous experimental protocols and inconsistent treatment of missingness mechanisms (MCAR, MAR, and MNAR) hinder systematic benchmarking across methods. This work investigates the robustness of Large Language Models for missing data imputation in tabular datasets using a zero-shot prompt engineering approach. To this end, we present a comprehensive benchmarking study comparing five widely used LLMs against six state-of-the-art imputation baselines. The experimental design evaluates these methods across 29 datasets (including nine synthetic datasets) under MCAR, MAR, and MNAR mechanisms, with missing rates of up to 20\%. The results demonstrate that leading LLMs, particularly Gemini 3.0 Flash and Claude 4.5 Sonnet, consistently achieve superior performance on real-world open-source datasets compared to traditional methods. However, this advantage appears to be closely tied to the models' prior exposure to domain-specific patterns learned during pre-training on internet-scale corpora. In contrast, on synthetic datasets, traditional methods such as MICE outperform LLMs, suggesting that LLM effectiveness is driven by semantic context rather than purely statistical reconstruction. Furthermore, we identify a clear trade-off: while LLMs excel in imputation quality, they incur significantly higher computational time and monetary costs. Overall, this study provides a large-scale comparative analysis, positioning LLMs as promising semantics-driven imputers for complex tabular data.

Problem

• missing data imputation
• Large Language Models
• tabular data
• missingness mechanisms
• benchmarking

Innovation

• Large Language Models
• Missing Data Imputation
• Zero-shot Prompting
• Tabular Data
• Benchmarking