Question

Data Mining (WEKA)

Summarize the main differences between the 3 runs of J48 in terms of changes

in accuracy, number of leaves in the decision tree, and how long it takes to build

the model.

Answer 1

3 runs of J48 in terms of changes in accuracy, number of leaves in the decision tree

1 Selects and open up a file segment1-challenge1.arff

2 Its uses J48 decision-tree-learner(DTL) ( Which typically,trees>J48)

3 Whatever supplied test try to set segment1‐test1.arff

4 Running accuracy is 96%

5 Evaluate on training set data is given by: accuracy of 99%

6 Evaluation of percentage of split is of 95% accuracy

7 J48 uses top‐down induction of decision trees

8 It is based on information theory

9 Proudced tree is easy to understand

10 Various  criteria for seleting attributes and very rare to make huge differences

how long does it takes to build the model.

I have tried to  run j48 in my system weka GUI for a dataset of around 27191 samples, 6 clases and each samples has 30 festures. I would say that time taken to build model is 24.28 seconds.
I wrote code snippet like this
             J48 cpcgod= new J48();
             long start-time = System.currentTime1();
cpcgod.build-Classifier(Multi-class-data1);
long stop-time1 = System.currentTime1();
long elapsed-time = stop-time1 - start-time;
The output elapsed time is given 4582 ms