TabArena
收藏资源简介:
TabArena是一个持续维护的表格数据机器学习模型基准测试系统。它通过手动挑选代表现实世界表格数据任务的51个数据集和16个表格机器学习模型,运行大规模的基准测试研究,以初始化一个公共排行榜。TabArena旨在提供一个可靠的评估环境,允许研究人员和从业者比较不同的机器学习模型。数据集经过精心挑选,以确保它们具有代表性,并且适用于独立同分布的表格分类和回归任务。TabArena的创建过程涉及模型的实现、超参数优化、数据集的挑选和评估设计。它旨在通过不断更新和改进来保持其相关性和可靠性,并通过开源项目的方式由社区共同维护。
TabArena is a continuously maintained benchmarking system for machine learning models on tabular data. It initializes a public leaderboard by conducting large-scale benchmarking studies with 51 manually selected datasets and 16 tabular machine learning models that represent real-world tabular data tasks. TabArena aims to provide a reliable evaluation environment that enables researchers and practitioners to compare different machine learning models. The datasets are carefully selected to ensure they are representative and suitable for independent and identically distributed (i.i.d.) tabular classification and regression tasks. The development process of TabArena involves model implementation, hyperparameter optimization, dataset selection, and evaluation design. It is designed to maintain its relevance and reliability through continuous updates and improvements, and is collaboratively maintained by the community as an open-source project.
数据集概述
基本信息
- 数据集名称:Spaces
- 平台:TabArena
- 点赞数:15
- 状态:Running
相关链接
- 数据集地址:https://tabarena.ai

- 1TabArena: A Living Benchmark for Machine Learning on Tabular DataAmazon Web Services, University of Freiburg, University of Mannheim, INRIA Paris, Ecole Normale Supérieure, PSL Research University, PriorLabs, ELLIS Institute Tübingen · 2025年



