Bayesian Embedding (BEMB)

Authors: Tianyu Du and Ayush Kanodia; PI: Susan Athey; Contact: tianyudu@stanford.edu

BEMB is a flexible, fast Bayesian embedding model for modelling choice problems. The bemb package is built upon the torch_choice library.

Installation

1. Install torch-choice following steps here.
2. Run the following script to install it.

   # Clone the repository to your local machine or server for tutorials.
   git clone "git@github.com:gsbDBI/bemb.git"
   # Install required dependencies.
   pip3 install -r requirements.txt
   # Install bemb from the Pip.
   pip3 install bemb
   # Check installation.
   python3 -c 'import torch_choice; print(bemb.__version__)'
   

Example Usage of BEMB

Here is a simulation exercise of using bemb.