Hi all
How can I deal with attributes ("not class") in the dataset that have values like (1/0, or Y/N) for classification problem?
Can " weka tool" deal with it directly without the need to convert it?
Hi Muna,
Your question is not so accurate as you need to identify whether the attributes you mentioned are regular attributes or class attributes. If those attribute are regular ones, converting them is not necessary as it depends on the algorithm or the the task itself. On the other hand, for classification purposes, if your class attribute has 0 or 1 values, then converting this class into a nominal type is required.
HTH.
Samer
In Weka, a binary attribute is simply declared in the header as a nominal attribute with 2 values Examples: Class {0, 1}, Married {Y, N}
There is a special approach for sparse files where you can not explicitly represent the 0 value.
More information at: http://www.cs.waikato.ac.nz/ml/weka/arff.html
Mr. Samer, thank you for your answer.
Indeed, my question is not about class but for attributes values which may have values like {Y/N, NIN, Y/Y N/Y}, how can I deal with? I can insert them directly into Weka tool or I must convert them to another format.
Mr. Eduardo Corrêa Gonçalves.
Thanks for your answer, but I am not talking about class values, I mean attributes " features" values which have values such as Y/N, N/Y, Y/Y ...etc
Ms. Muna,
Your question becomes clear now after editing process that you've done recently, and the clarification in your reply as well.
However, if you have such nominal attributes, then you can load them into WEKA directly.
Please let me know if you still have further issues.
HTH.
Samer
Mr. Ali, thanks for your answer, but I don't get it, can explain it more or can you send any tutorial or document for this issue, please.
I learnt weka tool from here, these lectures are very useful.
https://www.youtube.com/watch?v=Fg2x_zM3YTo&list=PLzVF1nAqI9VmC96TbvOPMkXToSmBMHJn7
- Badges
- Science topic
- Similar topics
- Computer Science
- Data Mining