openml

What it is and what it does

OpenML-Python is a client library that connects your Python environment to OpenML, an online platform for sharing and benchmarking machine learning datasets and experiments. It lets you programmatically fetch datasets, retrieve pre-defined machine learning tasks with standardized train-test splits, and access curated benchmarking suites—all without manually downloading files or managing data locally. The library wraps OpenML's REST API and integrates with pandas, scikit-learn, and numpy, so datasets come back in formats you already work with.

The package is designed for researchers and practitioners who want to run reproducible ML experiments on shared, versioned datasets, compare results across a community, or build on existing benchmarks. It handles the network communication, data parsing, and format conversion, so you focus on model development rather than data plumbing.

Use it for:

• Download standardized datasets for machine learning experiments without manual file management.
• Retrieve pre-defined classification or regression tasks with consistent train-test splits for benchmarking.
• Access curated benchmarking suites to run your models against a community-agreed set of problems.
• Upload experiment results back to OpenML to contribute to collaborative research and compare performance.
• Build reproducible ML pipelines that reference datasets by ID, ensuring consistency across runs and teams.

Worth the install?

AI-flagged interpretation of the facts on this page — verify before relying

OpenML-Python provides a Python interface to download and upload datasets, tasks, and machine learning experiment results from the OpenML online platform for collaborative open science research.

Yes. The package is actively maintained, has low install friction, carries a permissive BSD license, and solves a real problem for anyone doing collaborative or reproducible ML research. It integrates cleanly with the scientific Python stack (scikit-learn, pandas, numpy) and has no known vulnerabilities. Install it if you work with OpenML datasets or benchmarks; skip it if you manage datasets locally or use a different platform.

Install

openml on PyPI

pip install openml

uv add openml

poetry add openml

Installing openml

Before you install

Low friction installation with a pure-Python wheel. Active maintenance with a recent commit on 2026-08-10 and steady development since first release in 2018. Depends on well-established packages like scikit-learn, pandas, and requests.

License in practice

BSD 3-Clause License (permissive). You may use, modify, and distribute the package freely in commercial and private projects provided you retain the copyright notice and disclaimer.

Quickstart

pip install openml

import openml

# Download a dataset
dataset = openml.datasets.get_dataset("credit-g")
X, y, categorical_indicator, attribute_names = dataset.get_data(target="class")

# Or retrieve a task
task = openml.tasks.get_task(31)
train_indices, test_indices = task.get_train_test_split_indices(fold=0)

Requires Python 3.8 or later. Network access to openml.org is needed to download datasets and tasks.

Verify before relying

• Whether the package handles authentication or API keys for private datasets or uploads.
• Performance characteristics when working with very large datasets or many concurrent requests.
• Offline mode or caching strategy for repeated access to the same datasets.

Package facts

Evidence: openml-0.15.1-py3-none-any.whl

Tags

More Software Development packages