Installation Guide

We recommend installing HyperGBM with conda or pip. It’s also possible to install and use HyperGBM in a Docker container if you have a Docker environment.

As for software, Python version 3.6 or above is necessary to install HyperGBM.

Using Conda

Install HyperGBM with conda from the channel conda-forge:

conda install -c conda-forge hypergbm

Using Pip

Install HyperGBM with different pip options:

  • Typical installation:

pip install hypergbm

  • To run HyperGBM in JupyterLab/Jupyter notebook, install with command:

pip install hypergbm[notebook]

  • To support experiment visualization base on web, install with command:

pip install hypergbm[board]

  • To run HyperGBM in distributed Dask cluster, install with command:

pip install hypergbm[dask]

  • To support dataset with simplified Chinese in feature generation,

    • Install jieba package before running HyperGBM.

    • OR install with command:

pip install hypergbm[zhcn]

  • Install all above with one command:

pip install hypergbm[all]

Using Docker

It is possible to use HyperGBM in a Docker container. To do this, users can install HyperGBM with pip in the Dockerfile. We also publish a mirror image in Docker Hub which can be downloaded directly and includes the following components:

  • Python 3.8

  • HyperGBM and its dependent packages

  • JupyterLab

Docker image tag naming scheme:

  • <hypergbm_version>: Python + JupyterLab + HyperGBM + HyperGBM notebook plugins

  • <hypergbm_version>-cuda<cuda_version>-cuml<cuml_version>: above + CUDA toolkit + cuML

  • <hypergbm_version>-cuda<cuda_version>-cuda<cuml_version>-lgbmgpu: above + GPU enabled LightGBM

Download the docker image:

docker pull datacanvas/hypergbm

Run a docker container:

docker run -ti -e NotebookToken="your-token" -p 8888:8888 datacanvas/hypergbm

Then one can visit http://<your-ip>:8888 in the browser and type in the default token to start.

Requirements for GPU acceleration

  • cuML and cuDF

HyperGBM accelerates data processing with NVIDIA RAPIDS cuML and cuDF. Please install them before running HyperGBM on GPU. For detailed instructions, check the link Get RAPIDS.

  • LightGBM with GPU support

Default installation of LightGBM does not support GPU training. Please ensure LightGBM with GPU support before installing HyperGBM. For detailed instructions, check the link LightGBM GPU Tutorial.

  • XGBoost and CatBoost with GPU support

Default installations of XGBoost and CatBoost have supported GPU training. However, if you build them from source code, please enable GPU support.