LLMOverTab: Tabular data augmentation with language model-driven oversampling

Tokimasa Isomura, Ryotaro Shimizu, Masayuki Goto

Abstract: Highlights•Introduces LLMOverTab for oversampling in imbalanced tabular data.•Surpasses traditional methods like SMOTE and other LLM approaches.•Uses prompt engineering to generate meaningful synthetic instances.•Finds LLMOverTab excels especially with LLM prediction models.•Suggests exploring different LLM architectures and prompt techniques.