Estimating effort to port code from Matlab to Octave

I just read a fascinating paper: http://www.psy.cmu.edu/~ckemp/Papers/kempt08.pdf

In my opinion it takes the whole area of machine learning to a开发者_如何学JAVA completely new level because it flexibly discovers the structure of data (and doesn't only try to find a best fit for an existing structure).

The code is also available: http://www.psy.cmu.edu/~ckemp/code/formdiscovery.html

I tried to do a few experiments of my own - but unfortunately I don't possess Matlab. I tried it with Octave but it only produced all kinds of error messages, which I don't understand (I am no expert on these programs).

Could anybody perhaps have a quick look if these problems can be solved easily (or at all)? Perhaps the solution will be an easy one (this is my hope after all).

This would really be a big help! I am very much looking forward to trying a few data sets of my own.

---

I didn't look into the source code, but if you are going to convert it, these links might help:

• Porting programs from Matlab to Octave
• Differences between Octave and MATLAB

---

I was recently looking for an answer to the same question and found this old post. Just to add my two cents for anyone else looking... There is an Octave library called Missing Function Library that "Finds functions that are in Matlab but not in Octave". Also, there are a bunch of great packages (read "Toolboxes") for Octave as well on SourceForge. Hope this helps!