You are on page 1of 15

3/16/2015

Decision Tree Learning - I

Rule Based Learning


How far can we go?
• Many many rules
• Check redundancy
• Grouping of the rules
• Application of the rules

1
3/16/2015

2
3/16/2015

Entropy

Entropy

3
3/16/2015

4
3/16/2015

Entropy

5
3/16/2015

6
3/16/2015

Example: Play Tennis

The target
concept to
be learned

7
3/16/2015

8
3/16/2015

9
3/16/2015

Which Attribute is the Best Classifier?

10
3/16/2015

The Resulting Decision Tree

11
3/16/2015

Decision Tree for Play Tennis


Nodes specify
Possible values of test of some
the attributes attribute of the
instances

Classification according
to the target concept

12
3/16/2015

The ID3 Algorithm

13
3/16/2015

14
3/16/2015

Decision Tree Learning


• Among most popular of inductive inference algorithms
• A method for approximating discrete-valued target
function
• Learned function is represented by a decision tree
• The decision tree can also be interpreted as a set of ‘if-
then’ rules
• Ex:
– Diagnosis medical cases, assess credit risk of loan
applications

15

You might also like