CS513- Homework 5- Solved

 



 

Consider the data in Table above (end of chapter 6 or 8). The target variable is salary. Start by discretizing salary and age as follows:

 Less than $35,000                      Level 1

 $35,000 to less than $45,000   Level 2

 $45,000 to less than $55,000   Level 3

 Above $55,000                             Level 4

 

0   – 30                                               <= 30

31 - 40                                               <= 40

Above 40                                           <= 50

 

5.1 Construct a classification and regression tree to classify salary based on the other variables only one split level.

Hint: you may want to set up the excel file like the following

Split
PL
PR
Level
P( j |tL )
P( j |tR)
2PL PR
Q(s|t)
Φ(s|t)
1
0.273
0.727
L1
0.333
0.125
0.397
0.583
0.231
 
 
 
L2
0.333
0.250
 
 
 
 
 
 
L3
0.333
0.375
 
 
 
 
 
 
L4
0.000
0.250
 
 
 
2
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

5.2

The “breast cancer dataset” in CANVAS was obtained from the University of Wisconsin Hospitals, Madison from Dr. William H. Wolberg. The features in the dataset, described below, have been categorized from 1 to 10.

Use these categorized features to answer the following questions.

Important: make sure your categories are represented by the “factor” data type in R and DO NOT replace the missing values. 

 

     Features                      Domain

   -- -----------------------------------------

   Sample code number               id number

   F1. Clump Thickness               1 - 10

   F2. Uniformity of Cell Size       1 - 10

   F3. Uniformity of Cell Shape      1 - 10

   F4. Marginal Adhesion             1 - 10

   F5. Single Epithelial Cell Size   1 - 10

   F6. Bare Nuclei                   1 - 10

   F7. Bland Chromatin               1 - 10

   F8. Normal Nucleoli               1 - 10

   F9. Mitoses                       1 - 10

   Diagnosis Class:                 (2 for benign, 4 for malignant)

 

 

 

5.2

Use the CART methodology to develop a classification model for the Diagnosis.