EconML (Microsoft) — Heterogeneous Treatment Effects

What it does

Provides practical instructions and code examples for using Microsoft's EconML library to estimate heterogeneous treatment effects and causal effects from observational data. Covers Double ML (LinearDML), causal forests, and other orthogonal statistical learners with runnable Python snippets and references to official docs.

When to use it

Use this skill when you need to estimate average or conditional treatment effects from observational datasets where treatment effects vary across units, e.g. personalized policy evaluation, uplift modeling, or causal effect estimation in economics and marketing. It's useful for data scientists who want reproducible examples for Double ML and causal forest workflows.

What's included

• Scripts: example Python snippets demonstrating LinearDML and CausalForest usage (in-skill code samples).
• References: links to EconML documentation and GitHub.
• Instructions: installation note and step-by-step example code showing how to fit models and extract ATE or heterogeneous effects.

Compatible agents

Best used by agents and environments that can run Python snippets (agents with Python execution or code-running capability, e.g. Codex/Copilot-style runtimes, Claude Code).