-3

I'm interviewing for some quantitative researcher positions at some hedge funds, and I've been told that there will be one interview session focused on stats, and one focused on ML, among others. This made me realize that I have a hard time distinguishing between stats and ML because there's such a great deal of overlap, although I think I have some kind of idea of what parts of stats might not be typically apart of ML.

I took stats courses in college and high school, and never were the words "machine learning" mentioned in those courses. In those courses, I recall more of a focus on things (most of which I have forgotten) like hypothesis testing, basic probability and common distributions, confidence intervals, bar/box plots/histograms, and univariate regression.

What part of stats would you consider to not typically be part of ML and are important for practical data inference/analysis/prediction in the financial industry, particular at hedge funds / prop trading firms?

user5965026
  • 101
  • 1
  • 3
    Does this answer your question? How did you become a quant? – chrisaycock Jun 30 '20 at 04:02
  • Also, ML is a broad topic that has nothing to do with the stats you covered in high school. Try one of the many MOOCs or books on the subject. – chrisaycock Jun 30 '20 at 04:04
  • @chrisaycock Not really to the first question. I'm basically asking what kind of things are considered statistics, but not ML, by quants? The recruiters I've spoken with seems to distinguish between the two. I'm not very familiar with what constitutes as stats, but not ML, and I'm hoping to get clarification on that. – user5965026 Jun 30 '20 at 05:07
  • 1
    I think some of the answers to this question might steer you in the right direction: https://quant.stackexchange.com/questions/38862/what-are-the-quantitative-finance-books-that-we-should-all-have-in-our-shelves – amdopt Jun 30 '20 at 13:13

2 Answers2

4

To be honest this question isn't really within scope of QuantFinance because it is borderline opinion based but I will contribute an answer to the community regardless.

If you cannot determine the difference between stats and ML then you might consider that you do not understand the uses and methods used within each relevant discipline and may have to do more research, before being successful in an interview.

Statistics covers the mathematics of sampling and distributions of data. Hypothesis testing, simple probability and common distributions are so basic that you will rely on these constantly as a quant. I suggest you make every effort to remember your studies.

ML covers the variety of model techniques, and model fitting techniques, used to solve supervised or unsupervised problems, i.e. you have input data and you want to predict output data or perform some analysis on that data. That analysis may be statistical or it may be algorithmic. For example K-means clustering is not a "statistical technique", it is an algorithm.

Attack68
  • 10,176
  • 1
  • 17
  • 43
  • I wouldn't say that I can't determine the difference per se, but it's difficult for me to say "hypothesis testing and probability distributions belong to stats and not to ML" because that's not true. In ML, probability theory is very commonly used, and you also use other common statistical analysis techniques like hypothesis testing and confidence intervals (e.g., estimating the confidence intervals of OLS estimators). It's just when I spoke with the recruiter, it seems like they have a clear distinction between ML and stats, and I just don't really see that. – user5965026 Jun 30 '20 at 06:11
  • With that said, and similar to what you said, I feel that ML has more of a focus on algorithms. In my undergrad and high school level stats courses, the buzzword "ML" was never mentioned and everything was taught from a non-algorithmic perspective. Similarly, in the ML courses I've taken in grad school, traditional stats topics like hypothesis testing, p-values, confidence levels, were rarely referred to; however, The ELements of Statistical Learning (which I consider to be more of a ML book than stats) uses these topics frequently. – user5965026 Jun 30 '20 at 06:13
  • 1
    You use calculus and linear algebra a lot in ML as well, but I am assuming you can differentiate the two? They have uses outside of ML, as does statistical theory. If you draw a Venn diagram of "ML" and "Statistical theory" there will be 3 regions and you seem fixated only on the overlapping region and convinced that the overlap is largest enough to consider the two categories effectively synonymous. Try finding a use of statistical theory outside of ML, and some element of ML that does not use stats and perhaps all will become clearer from your perspective... – Attack68 Jun 30 '20 at 06:36
  • I don’t agree with this. Consider Frank Harrell’s Regression Modeling Strategies book. That book is about making accurate predictions, but Harrell would call himself a statistician, not a machine learning guy. (Indeed, his blog admonishes much of machine learning.) – Dave Dec 29 '21 at 15:38
  • OK. everyone is welcome to an opinion on a subjective matter. Frank Harrell is one person. – Attack68 Dec 29 '21 at 19:16
0

In broad, ML and statistics belong to data science. In a mean way, ML can be considered to be statistical models for prediction/classification problems. In machine learning, you will learn to deal with models like decision tree, svm, neural network, etc. When you apply these models to a particular problem, you will use basic statistical terms to evaluate the performance of models. What you learned prob&stats at college is essential to understand machine learning. You only need to learn a good course in machine learning, and I will suggest you to take the course of Prof. Andrew Ng. on Coursera. happy learning!

Thao Bui
  • 11
  • 2