PDA

View Full Version : Distinction between nominal and continous attributes in Weka



Dhafer
02-22-2007, 06:36 AM
Hello forum,


I’m new in this forum, I introduce myself:
I’m Dhafer Lahbib, master student. I’m working in automatic generation of classification rules especially in attribute regroupement in the rules antecedent. Of course, I’m interested to Weka, one of the best tools in the field of data-mining.

I have some questions.

Do someone know how do weka distinguish between nominal and continuous attributes ?
In the Weka source code is this done automatically or it is done by constructing a meta-data file through a kind spreadsheet?

thanks in advance,
Dhafer.

Mark
03-01-2007, 04:58 PM
Hi,

Weka will do its best to automatically determine the type of attributes. If your data is in an arff file, then the header of the file explicitly declares the type of each attribute. If you are loading data from a CSV file. then Weka will read data values and attempt to guess the correct type for each attribute. In the case of databases, various database types are mapped to Weka's attribute types.

Cheers,
Mark.