Installation

This page will guide you through the installation procedure of torch-choice and bemb.

There are two parts of this project: the torch_choice library consisting of data management modules, logit and nested-logit models for consumer choice modelling.

For researchers wish to use the Bayesian Embedding (BEMB) model, they need to install an additional bemb package, which was built on the top of torch_choice.

Note Since this project is still on its pre-release stage and subject to changes, we have not uploaded our packages to PIP or CONDA. Researchers need to install these packages from Github source code.

Option 1: Install using Source Code from Github Repository

To install torch_choice and bemb from source, 1. Clone the repositories of both torch_choice and bemb to your local machine. 2. Install required dependencies (e.g., PyTorch and PyTorch-Lightning). 3. For each of repositories, run python3 ./setup.py develop to add the package to your Python environment. 4. Check installation by running python3 -c "import torch_choice; print(torch_choice.__version__)". 5. Check installation by running python3 -c "import bemb; print(bemb.__version__)".

Option 2: Install using Pip

The torch-choice is available on PIP now here! You can use pip install torch-choice to install it. Note: We are working on publishing BEMB to PIP.