Description
1. (R and Python) Modify your program in Assignment #3 to do followings.
a. Prompt the user whether to run regression or classification.
b. If regression is chosen, perform the linear regression as you did in Assignment #2. (You have nothing
to work on the regression algorithm in this assignment).
c. If classification is chosen, ask the user the filename of the training and test dataset. (Assume the
column location of the class variable is the same for both training and test dataset.)
d. If classification is chosen, prompt the user to choose (i) LDA and (ii) QDA, or (iii) RDA.
e. Perform (i) LDA and (ii) QDA, or (iii) RDA depending on the choice by the user. Use a file named
“veh.dat” for the training and ‘vehtest.dat’ as the test data in this assignment.
f. For RDA, consider the extended model with two parameters. Try values in [0, 1] with increment of 0.05
for both the parameter values. Choose the parameters that minimize the test error rate. The program
must be able to choose the optimal parameters by itself after comparing test error rates. Also produce a
(3-dimensional) plot that displays both parameter values and their corresponding test accuracy. I.e. X1
is for parameter 1, X2 is for parameter 2, Y is for test accuracy. (Note: You need to estimate 𝜎𝜎”. Use
the average of diagonal elements of the pooled covariance matrix 𝑆𝑆$
“.)
g. The output file for classification generated by the program must look like below. (The numbers are
fictitious).
ID, Actual class, Resub pred
—————————–
1, 1, 1
2, 2, 2
3, 1, 1
(continue)
Confusion Matrix (Resubstitution)
———————————-
Predicted Class
1 2 3 4
Actual 1 239 14 6 8
Class 2 12 153 5 12
3 2 4 98 2
4 3 6 8 123
Model Summary (Resubstitution)
——————————
Overall accuracy = .793
ID, Actual class, Test pred
—————————–
1, 1, 1
2, 2, 2
3, 1, 1
(continue)
Confusion Matrix (Test)
———————————-
Predicted Class
1 2 3 4
Actual 1 239 14 6 8
Class 2 12 153 5 12
3 2 4 98 2
4 3 6 8 123
Model Summary (Test)
——————————
Overall accuracy = .793
