Respeced All,
I have KDD dataset 10% i am using it in WEKA. for all other algorithms it works well but for any tree algorithm it not gives any result can anyother hep me for this issue or someone already have KDD cup dataset 99 10% please send me. i will be grateful. i also attached the image file.
Hi,
You have half million instances, and the program says it is building the model (lower left corner), depending on your computer, I think it's going to take a while for building it. Moreover, since you are using the training data for testing, the program will perform a default ten-fold cross validation, which requieres more time than using an independent test set. What other models have you tried that worked fine?
Best,
Juan Carlos
You may also try to reserve a larger amount of memory for Weka. For instance, 2.0GB will be reserved if you use the below command to launch Weka
java -Xmx2048M -jar weka.jar
Select another attribute with two distinct values to use as a class and build the J48 tree then visualize it.
Regards
Correct and deceleration:
In the training option that you used (i.e., "Use training set"), WEKA performs resubstitution estimate (not cross-validation as mentioned above).
HTH.
Samer
- Badges
- Science topic