Teaching
Machine Learning with Social Data
“Machine Learning with Social Data: Opportunities and Challenges” is a joint-listed undergraduate course between Sociology and the Center for Statistics and Machine Learning. The course uses machine learning tools to study social data, with equal attention to the opportunities these tools create and the challenges that arise when the data represent real people and social systems.
The course is designed for two overlapping audiences. Social scientists and digital humanities scholars see how machine learning can help them learn about humans, make policy, and help people. Computer scientists and data scientists see how a social science research-design perspective can improve technical work and open up new applications for their skills.
Course Design
The course is organized around three thematic modules: the target task, the data source, and guiding values. Each module receives several weeks of attention, while recurring class meetings connect an applied example, an opportunity or challenge raised by the theme, and a representative machine learning technique.
Target Task
Measurement, prediction within a population, prediction in a new population, and causal inference as prediction to a counterfactual population.
Data Source
Experimental data, designed data such as surveys, and found data such as administrative records, digital traces, and electronic health records.
Guiding Values
Privacy, fairness, racial justice, interpretability, and accountability in algorithmic decision making.