F# as a Octave/Matlab replacement for Machine Learning

EDIT: As another version of the ml-class course has started, I've made the repository private

Back when I was in college, I took three different courses that dealt with subjects related to machine learning and data mining. Although I didn’t lose interest on those matters, my work has led me in a totally unrelated direction, so I haven’t exercised any of that knowledge in about eight years or so. A few weeks ago, I stumbled upon Stanford’s online class on Machine Learning and decided to enroll. I want to revive many of the things I have forgotten and try to put them into practice, as nowadays it’s very easy to access large amounts of interesting data from all kinds of online sources.

The programming exercises of this class are supposed to be done in Octave or Matlab, and while I understand the advantages of these tools, my past experience (where all the exercises and projects were done either with SAS or with Matlab) shows me that not using a general purpose programming languages doesn’t help a lot in turning academic exercises into real world programs. As professor Andrew Ng said in the introduction, one of the goals of the class is for us to put machine learning into practice in real world problems we care about, so I decided that I’ll implement all the algorithms and exercises in F#.

More...