FeRG-LLM automates feature engineering for tabular data by leveraging large language models to generate meaningful features with reasoning capabilities, reducing the need for human expertise.

• Creates high-quality features by generating both feature ideas and their rationales
• Produces human-readable explanations for each engineered feature
• Significantly improves model performance across various tabular datasets
• Reduces the time and expertise traditionally required for feature engineering

This innovation matters because it addresses a critical bottleneck in machine learning workflows, making advanced data analysis more accessible to organizations without specialized ML expertise.

FeRG-LLM: Feature Engineering by Reason Generation Large Language Models