Big Data for Public Policy

Many policy problems involve prediction. For example, a budget office might want to predict the number of applications for benefits payments next month, based on labor market conditions this month. This course provides a hands-on introduction to the "big data" techniques for making such predictions.

These techniques include:

• procuring big datasets, especially through web scraping or API interfaces, including social media data;
• pre-processing and dimension reduction of massive datasets for tractable computation;
• machine learning for predicting outcomes, including how to select and tune the model, evaluate model performance using held-out test data, and report results;
• interpreting machine learning model predictions to understand what is going on inside the black box;
• data visualization including interactive web apps.
  

Students will put these techniques to work on a course project using real-world data, to be designed and implemented in consultation with the instructors.