Introduction to Supervised Machine Learning for Social Scientists

Course Code

MACS 23005 91

Course Description

This course will cover the fundamentals of Supervised Machine Learning (SML). Students will be introduced to key prediction techniques like k-nearest neighbors, naive bayes classifier, regression (linear, logistic, non-linear and regularized regression models), decision trees, random forests and deep neural networks. Besides, students will learn about the importance of parameter optimization and also how to evaluate their SML model predictions using different types of cross-validation (Monte Carlo, leave-k-out, block) and with different metrics for classification problems (e.g. precision, recall, confusion matrices) or regression problems (e.g. Root-Mean Squared Error, Pearson or Spearman’s correlation).

The course ends with a brief and conceptual discussion of bias, fairness and trustworthiness in SML and what is the role of interpretable methods, so that students can know where to go next to learn the state-of-the-art discussions on SML. Note that the main focus of the course will be on having the students learn the critical concepts and develop deep intuitions about the techniques. To achieve that, we will cover some of the crucial mathematical foundations, but students will also have the chance to test and play with many techniques using the R programming language (which is assumed to be known already).

Instructor(s)

Fabricio Vasselai

¹ UChicago students can self-register.

² Visiting students and pre-college students apply through the same application.