4
$\begingroup$

I need to analyze the 100k MovieLens dataset for clustering with two algorithms of my choice, between the likes of k-means, agnes, diana, dbscan, and several others. What tools (like Rattle, or Weka) would be best suited to help me make some simple clustering analysis over this dataset?

$\endgroup$
1
  • $\begingroup$ Weka can be pretty slow. Try ELKI, it has very fast implementations (in particular when using indexes) and a much wider choice of algorithms. $\endgroup$ Jul 15, 2013 at 14:37

4 Answers 4

3
$\begingroup$

Try R. Here you have a list of clustering packages available.

$\endgroup$
0
2
$\begingroup$

Python will give you all the flexibility you need. With the NumPy and SciPy cluster module you have the tools you need, and the datatypes of NumPy give you a good insight in how much memory you will use.

$\endgroup$
5
  • $\begingroup$ Why the downvote? Is it not a good answer, or do you think it's not a good tool? $\endgroup$
    – Peter Smit
    Jul 22, 2010 at 17:06
  • $\begingroup$ I second the python recommendation here, especially if you have trouble fitting the data set into r. If you do go the python route, have a look at "Programming Collective Intelligence" by Toby Segaran $\endgroup$
    – PeterR
    Jul 22, 2010 at 18:29
  • $\begingroup$ What downvote? I'm not familiar with Python yet, but it seems promising, I'll take a look. $\endgroup$ Jul 23, 2010 at 11:32
  • $\begingroup$ @Victor, the first vote was a downvote (so it's now 2 upvotes/1 downvote). Maybe someone didn't like this tool. Happy I could help :) $\endgroup$
    – Peter Smit
    Jul 23, 2010 at 11:34
  • $\begingroup$ I'll give you +1 (so that you now have your 2 original upvotes :), just because I agree that with many data points, Python may be a good option; now the minor point, IMO, is that there're not so many clustering algorithms available in SciPy compared to R. $\endgroup$
    – chl
    Sep 13, 2010 at 6:16
1
$\begingroup$

Python's scikit-learn library (ml library built atop numpy and scipy) is quite popular, very well documented and has several clustering algorithms as described here.

$\endgroup$
2
  • $\begingroup$ You're 3 years late. $\endgroup$ Apr 14, 2013 at 15:46
  • 1
    $\begingroup$ For posterity :) $\endgroup$
    – beardc
    Apr 14, 2013 at 16:51
-1
$\begingroup$

can we use matlab for this work?I also need to cluster movielense dataset, my data is mixed of numeric and categoric dataset. what can I do?

$\endgroup$
1
  • 1
    $\begingroup$ Hi samira, and welcome to the site! Your post seems to be a new question rather than an answer to the question. I suggest that you delete this answer and post it either as a comment. See also the help pages of this site. $\endgroup$ Jul 15, 2013 at 11:50

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.