Filters - supervised versus unsupervised

[Lionel C.]

Hi,

I am not clear on why (and how) filters are differentiated between supervised and unsupervised filters. Could anyone please explain?

Thanks,

LC

[Ian Witten]

On 26/03/2014, at 12:32 pm, Lionel C. <lcaz...@gmail.com> wrote:

Hi,

I am not clear on why (and how) filters are differentiated between supervised and unsupervised filters. Could anyone please explain?

The simple answer is that unsupervised filters don’t take account of the class value, whereas supervised ones do

— but why would anyone want to do that?

But you need to be careful when using supervised filters

— careful?? — what do you mean?

In “More data mining with Weka” we will examine supervised filters in detail, and answer these questions — and more. In the meantime, my advice is to just use unsupervised filters.

cheers

ian

Thanks,

LC

--
You received this message because you are subscribed to the Google Groups "WekaMOOC-general" group.
To unsubscribe from this group and stop receiving emails from it, send an email to wekamooc-gener...@googlegroups.com.
To post to this group, send email to wekamooc...@googlegroups.com.
Visit this group at http://groups.google.com/group/wekamooc-general.
To view this discussion on the web, visit https://groups.google.com/d/msgid/wekamooc-general/8bf8dbd3-a1ec-450e-8fda-34dfdfeeaab4%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Lionel C.]

Dear Professor,

I thank you very much for your answer.

By the way, I thank you for having put this course together. I find it excellent and very engaging. The lecture on SVM was especially enlightening.

Cheers,

Lionel

Student from Bristol