You are on page 1of 8

Sample Exam Questions

1.           Consider discretizing a continuous attribute whose values are listed below:


3, 4, 5, 10, 20, 32, 43, 44, 46, 52, 59, 61
 
Which of the following number of bins is not possible for using equidepth
bins?
A.     2
B.     3
C.     4
D.     5
E.     6
F.     All of the above

2. (Decision Tree.) Consider the following data table where Play is a class attribute.   The training
data is shown below.
 
Humidity Outlook Windy Play
L N N T
L N Y T
H Y N T
H Y Y F
L Y N F
L Y Y T
H N N F
 
 
a. Build a decision tree that is at least 2 levels deep (the root is at level 1), using the
information gain measure for attribute selection.  Show all calculations. 
 
b. Suppose the last row’s value for Humidity is a missing value “?”. Describe two ways in
which you can handle the missing value “?”.  
 
c. Suppose that we have false positive cost = 3, and false negative cost = 4. For the following
decision tree, decide how to label each leaf node in order to minimize the cost at that
node. Explain your answers.

You might also like