Hi I am working on uncertain data , when I convert certain value such as in to uncertain interval value such [19,21) based on equation , the question is under which datatype I have to define this interval especially in arff file in weka ?
Thanks
As far as I understand, you do not know the variability in your intervals so you cannot treat them like a nominal variable. If this is the case, I would preprocess the data before feeding into WEKA. I would make two columns out of one showing lower and upper bounds. (e.g., interval=[19,21) would become interval_lower_bound=19, interval_upper_bound=21).
I hope this helps
Thank you so much Hamdi ,
Good idea , do you think it will work or I have to modify the classifier ?
I wouldn't touch the classifier because it would be an overkill. I believe splitting into two variables should work.
If my suggestion does not work, you have to come up with other ways to transform your data. One another way would be to create two variables: one hold the lower bound and one holds the interval size. (e.g., interval=[19,21) =>interval_lower_bound=19, interval_length=2).
- Badges
- Science topic