Hi,
I want to perform decision tree using j48, I have done following steps,
1) loaded training dataset->performed attrib selection using cfssubseteval and bfs to remove highly correlated attrib->classifier->trees->j48->test option chosen as 'use training set'->start
in test set, i have ensured, same attributes are present as that of the training file
2) then used 'set test file' option and loaded the test file and re-evaluated it on the generated model.
Following are my doubts:
1)Should i use Infogain attrib eval method instead of cfssubseteval ?
2) Any pre-requisites before applying J48 for constructing a tree?
Kindly guide me...thanks in advance
I want to perform decision tree using j48, I have done following steps,
1) loaded training dataset->performed attrib selection using cfssubseteval and bfs to remove highly correlated attrib->classifier->trees->j48->test option chosen as 'use training set'->start
in test set, i have ensured, same attributes are present as that of the training file
2) then used 'set test file' option and loaded the test file and re-evaluated it on the generated model.
Following are my doubts:
1)Should i use Infogain attrib eval method instead of cfssubseteval ?
2) Any pre-requisites before applying J48 for constructing a tree?
Kindly guide me...thanks in advance