Search Menu

Machine Learning: An Applied Econometric Approach by Jann Spiess

We are very happy to announce the following great opportunity to learn a new skill:

The Berlin Network of Labor Market Research (BeNA) and the Berlin School of Economics (BSE) with the support of Collaborative Research Center TRR 190 Rationality and Competition are happy to announce that Jann Spiess, Assistant Professor from Stanford Graduate School of Business, will give two courses  on Machine Learning: An Applied Econometric Approach on September 2-4 (BSE) and September 4-6 (BeNA).

The two classes overlap significantly and are not designed to be taken together. Because of the very interactive nature of the course, we can offer the course only to a limited number of participants. Please register for the BeNA course until August 20 by sending an email to

For further details, please see below or visit the BSE website.

Short course descriptions:
Machine learning has created many engineering break-throughs from real-time voice recognition to automatic categorization (and in some cases production) of news stories. What is particularly tantalizing though is that machine learning is, at its heart, an empirical tool. Given the similarity to tools we know, it is tempting to ask whether it is merely old (econometric) wine in a new (machine learning) bottle.In the courses, we will argue that it is not. Far from it, we will discuss how these tools can powerfully improve and expand on the kind of empirical work we tend to do. At the same time, we will discuss their limitations and how they fit into the “econometric toolbox”. At a high level, this class will address these three questions:
How does machine learning work? What can machine learning tools do that our current toolbox cannot? Where can machine learning be used to generate new research output?

We will cover standard machine learning techniques with a focus on supervised learning (such as regularized regression and methods based on decision trees). Towards the end of the class, we will also briefly discuss some unsupervised learning techniques (e.g. clustering). While the BSE Course (Sep 2-4) will focus on econometric underpinnings, the BeNA course (Sep 4-6) will focus on applications of machine learning in applied microeconomics like labor, health, and education.

Statistical packages make it trivial to implement machine learning in practice. But what makes them work? What statistical guarantees do they provide?
Further, where and how can machine learning be newly implemented in empirical microeconomics, specifically labor, health, and education?

The preliminary syllabus for each course:


Spiess_ML_BSE_Syllabus  Spiess_ML_BeNA_Syllabus