Formulating the machine learning problem, how research questions should be asked, and the trade-off of complexity versus accuracy (ARCHIVES S1E7)

We’re trying something new here, folks.  I’ve got 5 seasons and 60 brilliant episodes and I thought it would be fun, in the off season, to go back to the archives and highlight past conversations.

So using my trusty random number generator, I chose an episode at random.  So, we’re going back to 2018 to my conversation with John Alberg, co-founder of Euclidean Technologies, where machine learning is applied to the value investing problem.  

The part I’m highlighting starts around minute 20 and is about the formulation of the machine learning problem and how the research question should be asked.  I like this section because I think it really highlights how we can think about the tradeoff of degrees of complexity versus accuracy and the problem of overfitting.

Enjoy!