Obviously there are overlaps between econometrics and machine learning.

My question is: to what extent does doing graduate level econometrics actually overlap with machine learning? I mean this specifically in the sense: to what extent does doing graduate econometrics help you build either the specific concepts/methods, OR just build the general underlying math understanding that is useful for machine learning?

Another way of saying this would be: if you're interested in ML, but you can't take an ML course, to what extent is taking an econometrics course worth it?

A possible answers could be:

  • graduate level econometrics is for about 40% the same as machine learning, for another 30% it is different but still concerning underlying math that helps you understand ML, and for another 30% totally irrelevant. This is so because ....

  • if you do graduate level econometrics on topics X, Y, Z, you will only have to spend 20% of the time to learn ML topics U, W, V, compared to someone who didn't do the econometrics. This is so because...

By graduate level econometrics I mean stuff like: time series and panel data, estimators and test statistics for them (GMM, maximum likelihood, proofs of efficiency and consistency results, ...), both introductory and advanced graduate courses.


closed as primarily opinion-based by Alex R., Michael Chernick, Scortchi Apr 26 '18 at 6:52

Many good questions generate some degree of opinion based on expert experience, but answers to this question will tend to be almost entirely based on opinions, rather than facts, references, or specific expertise. If this question can be reworded to fit the rules in the help center, please edit the question.

  • 1
    $\begingroup$ " but you can't take an ML course, to what extent is taking an econometrics course worth it?" - how on earth can you not take an ML course in this day and age? Websites offering comprehensive online ML courses are second only to porn websites on the net these days. $\endgroup$ – Skander H. Apr 26 '18 at 5:56
  • $\begingroup$ This question is not really answerable. Machine learning is broad, as is economics. So skills in economics are a proxy for skills in machine learning. Reducing it to a percentage is vacuous. You could work on machine learning applications to economics, and that would constitute 100% of your time. Finally, a basic rule of machine learning is that if you want to optimize for something, don't pick a proxy, use the main KPI, in this case, learning machine learning. $\endgroup$ – Alex R. Apr 26 '18 at 5:59
  • 1
    $\begingroup$ The answer is: "Depends on the course." Some lecturers go in depth into programming and models, others teach basic statistics. ML is a broad field and as long as you work with statistics in the course, you can consider it "the right direction". $\endgroup$ – Nikolas Rieble Apr 26 '18 at 8:56
  • $\begingroup$ I think the emphasis in econometrics (particularly on the micro side) is on causal inference, whereas the emphasis in ML is on predictive /classification performance. There is very little overlap in tools and goals and techniques. You can take a look at Athey-Imbens 2018 AEA lectures for some recent developments in this area. The econometrics-stats connection (rather than CS version of ML) can be a bit closer. $\endgroup$ – Dimitriy V. Masterov May 1 '18 at 21:08