1
$\begingroup$

I am doing assignment of Data mining , and stuck on one thing book i am following is "introduction to data mining" by viper kumar , it is "How many instances are misclassified in given decision tree?"

Tree:

     True---------+ A+---------------False
     |                               |
     |                               |
     |                               +
+----+------+              True-----+B+-------False
|           |              |                  |
|  +(25)    |              |                  |
|           |              +                  |
|           |              C               +--+----------+
+-----------+              +               |             |
                           |               |  -(30)      |
                 True------+------False    |             |
                 |                |        +-------------+
                 |                |
                 |                |
          +------+--+          +--+-------+
          |         |          |          |
          | -(20)   |          | +(25)    |
          |         |          |          |
          +---------+          +----------+

i am wondering how to calculate it.

Note: i am not asking for solution rather the concept behind it.

$\endgroup$
1
  • $\begingroup$ Welcome to CV. Rather than providing a link and leaving it up to the readers to figure your precise question out, it would be much better to spell it out in your statement. Please revise your query to reflect this. $\endgroup$ Mar 29, 2016 at 13:56

1 Answer 1

1
$\begingroup$

I guess you are getting confused because you've build the perfect decision tree for the data, thus it does not have any misclassification error at all. However, the exercise is asking you to reflect "on the greedy nature of the decision tree induction algorithm".

The point is: you build the tree, and first you decide which attribute (A, B or C) is going to be the root. You do it by looking for the attribute that would produce the best misclassification error on its own. However, this might lead you to a tree that is not ideal (i.e. it will force you to take an attribute which is not the ideal as a root of your tree, when you will later look at the result of the complete tree). That's the drawback of the greedy nature of the algorithm.

Hope it is clear and it helps you, still not providing the solution.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.