Machine Learning for Social Science — Short Course

Materials for a short course introducing machine learning and AI methods to social scientists: prediction and the ML toolkit (Day 1), and machine learning for causal inference (Day 2).

Contents

• ML Primer.pdf — Day 1 slides.
• ML and Causal Inference.pdf — Day 2 slides.
• Python code/, R Code/, Stata Code/ — code reproducing the examples in each language. Most examples appear in all three; some are single-language. Each directory has a run_all script.
• Data/ — the baseline (source) data the examples read. Larger intermediate outputs are regenerated by running the code.
• map of code to material.xlsx — which code files correspond to which section of the notes.
• Note on Software for Lectures.txt — start here to set up Python (via Anaconda), Stata, and/or R.

Getting started

1. Read Note on Software for Lectures.txt.
2. Create the shared Python environment: conda env create -f environment.yml.
3. Run the code in your language of choice.

All three languages share one Python environment (mlssshort); Stata's Python integration and R's reticulate both call into it. See the software note for details.