One question that definitely has plagued me for a while is that of evaluation schemes of recommender systems... On multiple occasions I have designed recommender systems that give a fairly good user experience, but I just have not been able to quantify these results... any pointers on that will be really appreciated.
Pretty interesting stuff..
Personally I wasn't a big fan of recommenders that rely on explicit interaction with the end users simply because it may be too much work for them.
Hunch, on the other hand seems to frame questions in a much more entertaining manner, I could see my colleagues having fun while answering questions, and even after 20, they went ahead and answered a few more.
The whole process definitely makes me think again about how to go about approaching recommender systems...
For what it's worth, I think it's important to start with interesting problems. My first interaction with ML was an implementation of Naive Baye's for classifying spam, borrowing much ideas from PG's A Plan for Spam from scratch (ie no libraries). This is what got me really interested in the field, much more than randomly picking up topics- there are just so many areas to choose from. Another approach would be to read up on standard supervised learning techniques and just observing how the parameters for these algos behave on datasets. something like a Weka really comes in handy if you wish to focus on analyzing behavior of such techniques first. Best of luck!
Second Michael's opinion...
Some really good python libraries out there besides NumPy and SciPy.
Pandas comes to mind http://code.google.com/p/pandas/
Pretty good for data analysis.
Besides this, I think there was a talk on Python for the Finance world in this year's PyCon, possibly worth a look too...
I don't how accurate it is to say they are not of use any more... agreed SVM's seem to be particularly good. But as far as I know, NN's seem to pretty well for handling data like signals...
I think besides from being computationally expensive, the fact that there is very little possible analysis for linking the outputs with the inputs and their nature probably works against them.
well so here is what I am doing instead...
I get only a notion of the likes of the user, and since "everything else" is too huge a domain to consider as dislike, I try and rank the user's like concepts and pick the lower one's as the one's they like lesser. This only for initialization, and then I leave the filter to tune itself through feedback. I know it's not the best approach in the world, but let me see how this shapes up. Thanks of course, and any other ideas still welcome!
well there are loads of lectures on videolectures.net
definitely worth a view, in particular those from machine learning summer school(primarily because audio quality tends to be pretty sucky in some of the other ones)...
Looks pretty interesting...
however a small UI suggestion.... it seems to occupy a fair amount of the available screen space.
How about something like a "bitcrumb powered searchbox" in the browser itself as opposed to a panel on the left??
nice post... while I do agree with most of the content on your post...could you elaborate a little more on the approach you guys are planning to take in particular?
Also, one thing that probably may work against startups is while they may be really good at reacting to issues, at the same time, they are probably more prone to them to begin with.. I'd love to hear what others from HN have to say about this...
Good pointers, yes Andrew Ng seems really amazing that way, I had actually seen his videos too, forgot to post that link..
Thanks guys for these suggestions.
i did try the libsvm implementation in weka.. but it wasn't available for all datasets (my guess being it works only for binary classification), ended up using the SMO algorithm which gave pretty good results too...
great resource!! what is it about NZ and machine learning :)
(weka comes from there, for those who are not aware)
Anyway, PyML (http://pyml.sourceforge.net/) and biopython (http://biopython.org) are two other packages I found pretty useful and can be explored..
great replies guys... some very good pointers, especially ventureloop was a great find for me.
thanks for the inputs again.
probably another area for others interested in the field are domains on finance (like imurray suggested). i found
http://www.cerebellumcapital.com/ pretty cool.
but no internships, only jobs.