Simulation Study: Multicategory Partial Identification Learning

This repository contains code and notebooks for simulation for the paper Multicategory Partial Identification Learning for Individualized Treatment Rules.

Project Structure

• helper.py — Utility functions for simulations
• mcat_learn.py — Main learning algorithms and methods
• simluation.ipynb — Main simulation notebook
• pixi.toml — Pixi environment configuration

Environment Setup

This project uses Pixi for reproducible environment management. To get started:

1. Install Pixi: https://pixi.prefix.dev/latest/
2. Run pixi install in the project directory to set up the environment.

Usage

• Run the simulation notebooks in your Jupyter environment and select the pixi kernel (which is located in you project directory under .pixi/envs/default/bin/python if not automatically selected, see documentation).