0
$\begingroup$

We are trying to run J48 on a classified data set. Our class attribute has two possible values ( 0,1) when running J48 the tree terminates at the very first node and doesnt process any further.

Instead of considering (0- false) as the starting point of J48. How can we consider running J48 by selecting (1-true) as the starting point of the tree?

Any suggestion will be greatly appreciated.

$\endgroup$
1
  • 1
    $\begingroup$ This might occur if J48 decides it cannot find meaningful nodes. What is the a priori distribution of the classes and how high is your confidence parameter (C)? $\endgroup$ – S van Balen Jan 4 '17 at 13:51
1
$\begingroup$

I am assuming that you have only 1 attribute (numeric). What you can do is

  1. modify the .arff file and set 0s to 1 and 1s to 0.
  2. Then you can again run j48 classifier and visualise the decision tree.

It would give you inverted result. Also, try watching this video.WEKA on MOOC

$\endgroup$
0

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.