Tutorials: Practical Guidelines

• For the tutorial sessions at the IMA 2026 World Congress, we expect participants to bring their own laptops.
• For some tutorials, the installation of some software will be necessary.
• To avoid wasting time on installation problems during the tutorial session, we ask all participants to install all the necessary software on their laptop prior to the tutorial session.
• I you encounter difficulties with the installation, you are invited to directly contact the organizer of the tutorial session.

Introduction to EUROLAB (Hannes Serruys)

• Wednesday 1 July, 14:15–15:45, Room F (2300)
• Contact: hannes.serruys@ec.europa.eu

Please note: this is still a demo version of EUROLAB.

Requirements

EUROLAB currently runs on Windows only¹, and a minimum of 16 GB of RAM is required.

Setup

1. Install EUROMOD. Download and install the latest version of the EUROMOD software from the JRC download page.

2. Install EUROLAB. Download the EUROLAB installer from the shared drive and run it (no administrator rights are required).

   The file can be trusted. Because of its size, Google Drive cannot scan it for viruses and will warn you when downloading — choose Download anyway. Likewise, Windows may flag the unsigned installer when you run it; click More info and then Run anyway.

3. Get it running beforehand. Launch EUROLAB and make sure it starts up correctly ahead of time, so any setup issues can be sorted out before you need it.

What comes next

In the next few days a EUROMOD model will be added to the same Google Drive folder. You will need to download it, but it requires no additional installation.

Support

If you encounter any problems, please email Hannes Serruys at hannes.serruys@ec.europa.eu.

Policy modelling in EUROMOD with Generative AI (Hugo Cruces)

• Wednesday 1 July, 16:30–18:00, Room F (2300)
• Contact: hugo.cruces@ec.europa.eu

AI Coding Workshop — Setup Instructions

In this session we are going to work with Visual Studio Code, Claude Code, and EUROMOD for AI-assisted coding and policy microsimulation. Before arriving to the workshop, please make sure you have completed the following setup steps.

1. Prerequisites

1. An Anthropic account: Create one at Anthropic.
2. A Claude subscription or API access: Claude Code requires a Pro, Max, Team, Enterprise, or Console/API account. At the end of this document there are alternative AI tools that will also work fine, in case you prefer those.
3. (Optional but recommended) Install Git for improved terminal compatibility with Claude Code.

2. Install Visual Studio Code (VS Code)

If you do not already have VS Code installed:

1. Download and install it from Visual Studio Code.
2. Open VS Code once installation is complete.

3. Install Claude Code

Claude Code runs in the terminal and integrates well with VS Code. Windows (PowerShell) Open PowerShell and run: irm https://claude.ai/install.ps1 | iex macOS / Linux Open a terminal and run: curl -fsSL https://claude.ai/install.sh | bash After installation, restart your terminal.

4. Authenticate Claude Code

1. Open a terminal inside VS Code (Terminal → New Terminal).
2. Run: claude
3. Follow the login instructions in your browser to authenticate your Anthropic account.

5. Verify the Installation

Inside the VS Code terminal, run: claude –version If a version number appears, the installation was successful.

6. Install EUROMOD

EUROMOD is the tax-benefit microsimulation model we will use during the workshop.

1. Go to the official download page: EUROMOD Download Page
2. Download the latest 64-bit installer for Windows.
3. Run the installer and follow the default installation steps.
4. Launch EUROMOD after installation to verify that it opens correctly. Notes: EUROMOD currently runs natively on Windows. Mac users typically run it through a Windows virtual machine. The software and policy models can be downloaded directly, but access to EU-SILC input microdata requires a separate request procedure.

7. Install Python and Set Up a Virtual Environment

Python is used by AI agents for data manipulation, scripting, and advanced analysis in combination with microsimulation models.

1. Download and Install Python
   1. Go to the official website: Python Downloads
   2. Download the latest stable version of Python 3 (e.g., Python 3.11 or newer).
   3. During the installation on Windows, make sure to check the box that says “Add python.exe to PATH” before clicking Install.
2. Verify the Installation Open a new terminal in VS Code and run: python –version (On some macOS/Linux systems, you might need to use python3 –version instead). Ensure pip (Python’s package installer) is also available: pip –version
3. Use Virtual Environments (Recommended) To keep your workshop dependencies isolated, it is best practice to use a virtual environment:
   1. Navigate to your project folder in the terminal.
   2. Create a virtual environment named .venv: Windows: python -m venv .venv macOS / Linux: python3 -m venv .venv
   3. Activate the virtual environment: Windows (PowerShell): .venv\Scripts\Activate.ps1 macOS / Linux: source .venv/bin/activate 3 / 4 IMA_AI_workshop_setup_instructions.md 2026-06-08
   4. Once activated, your terminal prompt will show (.venv). You can now safely install required libraries.

8. Alternative AI Coding Tools (Optional)

There are several alternatives to Claude Code that also integrate well with VS Code: GitHub Copilot — AI pair programming directly inside VS Code. OpenCode — Open-source terminal-based AI coding assistant. Cursor — AI-first code editor based on VS Code. Continue — Open-source AI coding extension for VS Code. Cline — Agentic coding assistant inside VS Code. We will focus on Claude Code during the workshop, but the main concepts transfer across all tools.

Decision modelling in Python: an introduction to the MISCore discrete-event microsimulation package (Koen de Nijs)

• Thursday 2 July, 10:30–12:00, Room F (2300)
• Contact: l.vanduuren@erasmusmc.nl

Installation instructions

We will work with an online binder environment, so no prep by participants is necessary. They can view the documentation of our python package on miscore.io; but to just have a laptop with internet access will be sufficient.

Tutorial session: Analysing tax-benefit reform impacts with PolicyEngine (Max Ghenis)

• Friday 3 July, 10:30–12:00, Room F (2300)
• Contact: max@policyengine.org

Installation instructions

questions about this tutorial or setup: max@policyengine.org

TL;DR — there is almost nothing to install

This tutorial is browser-based. If you bring a laptop with a modern web browser and can get online, you are ready. Everything core runs at policyengine.org with no account and no installation.

The optional steps below (a Google account for the notebook; a local Python install; the AI assistant) only matter if you’d like to follow specific parts hands-on. Skip them and you’ll still be able to do everything in the browser.

A note on countries: the hands-on examples use the US model, whose microdata is fully open. The UK model works identically, but its population microdata is not publicly downloadable, so UK population analysis is shown by the presenter rather than run on your own machine.

1. Required — for everyone (2 minutes)

1. A laptop with a current browser (Chrome, Firefox, Edge, or Safari).
2. Confirm you can reach policyengine.org and app.policyengine.org on the conference Wi-Fi (Eduroam).
3. (Optional) Create a free PolicyEngine account if you’d like to save your work — not required to follow along.

That’s it. The household calculator, policy reforms, and population-level impact analysis (budgetary cost, distribution, poverty, winners and losers) all run in the browser.

2. Recommended — to follow the Python notebook (5 minutes)

Part of the session uses a short Google Colab notebook, which runs Python entirely in your browser — nothing is installed on your machine.

1. Make sure you can sign in to a Google account (Colab requires one).
2. Open the tutorial notebook — from the tutorial hub at policyengine.org/tutorial, or directly: https://colab.research.google.com/github/PolicyEngine/tutorial/blob/main/notebooks/policyengine_tutorial_ima2026.ipynb
3. (Optional, to save time) Open it before the session and run the first cell — it installs the policyengine package (~2–3 minutes the first time).

If you prefer not to use Colab, you can simply watch this portion; the same results are reproduced in the web app.

3. Optional — to run locally instead of Colab (advanced)

If you’d rather run on your own machine, you need Python 3.10 or newer, then:

pip install policyengine

This installs the unified PolicyEngine package (both US and UK tax-benefit models). Quick check:

import policyengine as pe
result = pe.us.calculate_household(
    people=[{"age": 40, "employment_income": 50_000}],
    year=2025,
)
print(result.household.household_net_income)

Population-level microsimulation on the full national dataset is memory-intensive (it runs over ~100,000+ representative households). For the live session we use the web app or a small calibrated sample for that part; full-scale runs are better done after the workshop.

4. Optional — to try the AI assistant (demo)

PolicyEngine ships a Claude Code plugin that drives all of the above from plain-language prompts (e.g. “What’s the poverty effect of raising the Child Tax Credit to $3,000?”). In the session this is a demonstration — you do not need to install anything to follow it.

If you’d like to try it yourself afterwards, you’ll need Claude Code and an Anthropic account; setup instructions will be linked from policyengine.org/tutorial.

Summary

Questions about this tutorial or the setup: contact Max Ghenis directly at max@policyengine.org.