You are on page 1of 7

AAYUSH BADADE

BTECH CSIT A1
2018BTCS002
Data Warehousing and
Mining

PRACTICAL ASSIGNMENT
ON CLASSIFICATION
WEKA
Using the open source WEKA tool, perform Decision Tree Classification for a 2-class classification
problem for the "Diabetes" cases using 10-folds Cross-validation. Diabetes dataset is available in
the data folder in Weka.
Questions:
1. (a) Present the results of applying Decision Tree classification in the form of rules or tree.
(b) Apply Attribute Selection Filter to the dataset and then apply Decision Tree Classification.
2. Compare the results produced by both the approaches.
Do you find any difference in correctly identified instances and rule generated?
1.
(a) Decision Tree without Filter
(b) Decision Tree with Attribute Selection Filter
2.
 Comparing the results of both approaches one can see the Filter only considers data through
which most information can be gained i.e. only attributes of utmost necessity in determining
the result are considered. The rest is ignored.
 Coming to the results one can observe that the tree generated is smaller with lesser
components. Tree(a) has 20 leaves and its size is 39 whereas Tree(b) has 15 leaves and its
size is 29.
 It also takes less time to generate Tree(b).
 Coming to the accuracy of results:
Tree(a) has Correctly classified instances by 73.8281 % whereas Tree(b) correctly classifies
instances by 74.8698%. Thus, providing us with a more accurate Tree which can make
better predictions.

You might also like