15 attributes. It took 0.89 second to build the model
and the model generated a tree with a size of 473
and 323 leaves.
On the second scenario the algorithm was
run on a full training set containing 7,339 instances
with selected 8 attributes. It took 0.36 second to
build the model and the model generated smaller
and less complex tree with a size of 126 and 71
leaves making it less complex and faster than the
experiment conducted on all attributes.
In the first experiment I evaluated the
performance of J48 classifier unpruned tree in
predicting heart disease. The result of which is
given in table 1.1 below and their detail performance
measures used is depicted in table 1.2.
In the first case the algorithm was run on
a full training set containing 7,339 instances with
15 attributes. It took 0.89 second to build the model
and the model generated a tree with a size of 473
and 323 leaves.
In the second case situation the algorithm
was run on a full training set containing 7,339
instances with selected 8 attributes. It took 0.36
second to build the model and the model generated
smaller and less complex tree with a size of 126
and 71 leaves making it less complex and faster
than the experiment conducted on all attributes.