Heller Course Highlight: Causal Inference and Machine Learning in Studies Using Observational Data

Brielle Ruscitti, MS GHPM/MA SID '24

February 20, 2024

Brielle RuscittiHappy February! The spring semester is underway at the Heller School, we have settled into our classes and routines, and the days are beginning to get longer. Today I am going to highlight one of the classes that I am taking this spring – despite the long and semi-confusing name, is an interesting course taught by William Crown. Since I had some extra space in my schedule and it is my last semester at the Heller school, I sought out a course in the PhD program to complement what I have learned in some of my M.S. data analytics and econometrics courses to learn more about the emerging applications of machine learning and causal inference. 

This course is broken down into three main components – we typically spend the first bit of class discussing an article on the applications on causal inference or other important literature. As a class, we break down questions we had after reading and discuss the methods of the paper. This opens the class up and we are able to share our individual takes on the paper.

Following this short discussion, we break down new concepts, methods, and the background theory. This is the typical lecture portion of the class and allows us to understand the different methods before using them ourselves. For example, this past week, we discussed Directed Acyclic Graphs, which are an interesting way to draw out research questions and the potential relationships between variables. We made and presented our own directed acyclic graphs based on our own research question to share with the class and through discussion got feedback on additional variables to add and any blind spots. 

The last portion of the class is taught as an active learning lab session. Typically, we use a prepared dataset and apply the methods or techniques discussed in the lecture portion of the class to learn how to use R studio programming software to actually test out the methods ourselves. For example, we used a directed acyclic graph generator online that we then were able to code into R studio to create a directed acyclic graph. 

Overall, this class combines theoretical knowledge and practical applications to machine learning techniques that I will be able to use well beyond my time at Heller. I’m glad I branched out to test other program classes at Heller – would you take a machine learning class?