Machine Learning for Social Science

Aim

After completion of the course, the student should at an advanced level be able to:

• Use concepts and methods in machine learning to formulate and solve practical problems relevant for social scientific research.
• Use statistical software to estimate machine learning models, evaluate their fit, select their parameters, and interpret their results.
• Describe the strengths and weaknesses of different machine learning methods.
• Describe which machine learning models are suitable for specific applications/settings.
• Compare machine learning tools with traditional statistical methods, identify problems most suitably modeled with machine learning, and problems in which the two can complement each other.

Content

This course provides an overview of the key concepts and tools of machine learning (ML) that are relevant to social science research. First, a general introduction to ML is provided, where foundational ideas are reviewed and contrasted to those of traditional statistics. Then, central techniques in supervised learning (e.g., decision trees) and unsupervised learning (e.g., k-means) are introduced. In computer labs, students learn how to use these techniques in statistical software to solve practical problems relevant for social scientific research. Finally, the intersection between ML and causal inference will be considered.