Installation Guide

1 System Requirements

Installation Requirements:

  • Python >= 3.6 (3.7 is recommended!)

Supported Systems:

  • Linux (Ubuntu, …)

  • macOS

  • Windows

2 Preparations before Installation

We STRONGLY suggest you to create a Python environment via Anaconda:

conda create -n openbox3.7 python=3.7
conda activate openbox3.7

Then we recommend you to update your pip and setuptools as follows:

pip install pip setuptools --upgrade

3 Install OpenBox

3.1 Installation from PyPI

To install OpenBox from PyPI, simply run the following command:

pip install openbox

3.2 Manual Installation from Source

To install OpenBox using the source code, please run the following commands:

For Python >= 3.7:

git clone && cd open-box
cat requirements/main.txt | xargs -n 1 -L 1 pip install
python install

For Python == 3.6:

git clone && cd open-box
cat requirements/main_py36.txt | xargs -n 1 -L 1 pip install
python install

3.3 Test for Installation

You can run the following code to test your installation:

from openbox import run_test

if __name__ == '__main__':

If successful, you will receive the following message:

===== Congratulations! All trials succeeded. =====

If you encountered any problem during installation, please refer to the Trouble Shooting section.

4 Installation for Advanced Usage (Optional)

To use advanced features such as pyrfr (probabilistic random forest) surrogate and get hyper-parameter importance from history, please refer to Pyrfr Installation Guide to install pyrfr.

5 Trouble Shooting

If you encounter problems not listed below, please File an issue on GitHub or email us via


  • ‘Error: [WinError 5] Access denied’. Please open the command prompt with administrative privileges or append --user to the command line.

  • ‘ERROR: Failed building wheel for ConfigSpace’. Please refer to tips.

  • For Windows users who have trouble installing lazy_import, please refer to tips. (Deprecated in 0.7.10)


  • For macOS users who have trouble installing pyrfr, please refer to tips.

  • For macOS users who have trouble building scikit-learn, this documentation might help.