| Autor: |
Pietruszka, Michał, Borchmann, Łukasz, Jędrosz, Aleksander, Morawiecki, Paweł |
| Rok vydání: |
2024 |
| Předmět: |
|
| Druh dokumentu: |
Working Paper |
| Popis: |
We present a benchmark for large language models designed to tackle one of the most knowledge-intensive tasks in data science: writing feature engineering code, which requires domain knowledge in addition to a deep understanding of the underlying problem and data structure. The model is provided with a dataset description in a prompt and asked to generate code transforming it. The evaluation score is derived from the improvement achieved by an XGBoost model fit on the modified dataset compared to the original data. By an extensive evaluation of state-of-the-art models and comparison to well-established benchmarks, we demonstrate that the FeatEng of our proposal can cheaply and efficiently assess the broad capabilities of LLMs, in contrast to the existing methods. |
| Databáze: |
arXiv |
| Externí odkaz: |
|
