[prev in list] [next in list] [prev in thread] [next in thread] 

List:       wekalist
Subject:    Re: [Wekalist] Weka and Products database
From:       Peter Reutemann <fracpete () waikato ! ac ! nz>
Date:       2006-06-29 20:19:08
Message-ID: 44A435BC.2050806 () waikato ! ac ! nz
[Download RAW message or body]

> But i still have a question about Weka and the particularity of my DB.
> In your opinion, is Weka a suitable choice to analyze my entries?
> Because i can see its utility in many kinds of DBs, but i fail to see 
> where discretizing dates, product codes (unfortunately you can't see the 
> category that the item belongs to from its code) and number of items 
> bought leads to. It's just a big collection of orders and the products 
> they contain.

Why don't you modify your statement then? Instead of returning only the 
id of an item/order/etc, return the actual value from the referenced 
table (via a join). Then you could have the category, etc. for the 
items/orders/etc., which would be quite helpful for the classification 
process, I assume.
You could also try RELAGGS, a tool I integrated in my Proper framework, 
to work on relational databases (it's best to grab the source code from 
CVS!):
   http://www.cs.waikato.ac.nz/~ml/proper/

> I was looking at algorithms like Slope One and collaborative filtering 
> in general, and i'm still unsure witch choice i should adopt.

Sorry, can't help you there since I'm not familiar with those approaches...

Cheers, Peter
-- 
Peter Reutemann, Dept. of Computer Science, University of Waikato, NZ
http://www.cs.waikato.ac.nz/~fracpete/     +64 (7) 838-4466 Ext. 5174

_______________________________________________
Wekalist mailing list
Wekalist@list.scms.waikato.ac.nz
https://list.scms.waikato.ac.nz/mailman/listinfo/wekalist
[prev in list] [next in list] [prev in thread] [next in thread]