LOYOLA COLLEGE (AUTONOMOUS), CHENNAI – 600 034

M.Sc. DEGREE EXAMINATION – STATISTICS

YB 34

FIRST SEMESTER – April 2009

ST 1811 – APPLIED REGRESSION ANALYSIS

 

 

 

Date & Time: 30/04/2009 / 1:00 – 4:00 Dept. No.                                                   Max. : 100 Marks

 

 

SECTION – A

Answer All questions.                                                                           (10 x 2 = 20 marks)

1. What is a multiple linear regression model ?

2.Why do regressions have negative signs. Give reasons.

3. Explain BLUE.
4. Explain the co-efficient of determination
5. State any two ways in which ‘specification Error’ occurs.
6. What is multi collinearity?

7.What is the formula for finding the adjusted r-square?

8. What is Residuals ?
9. Why do we use Dummy variables in a model?
10. What are response and explanatory variables?

SECTION – B

Answer any Five questions. Each carries 8 marks.                             (5 x 8 = 40 marks)

11. What are the three components specified in a generalized linear model? Explain in detail.
12. Explain in detail categorical data analysis with examples. What are the two primary types of scales of categorical variables? Give example.
13. What is the form of logistic regression model? Also give link function for a logistic regression model?
14. Explain the four methods of scaling the Residuals ?
15. Write short notes on Residual Plot.
16. Estimate bo , b1, and s of  simple linear regression model by MLE
17. Give an application scenario to illustrate the simple regression model .
18. Write a short note on detecting multi collinearity.

SECTION –C

Answer any TWO questions. Each carries 20 marks.                         (2 x 20 = 40 marks)

19. Give an illustration and explain the following in detail :

a)Binomial logit models for binary data

b)Poisson log linear model for count data                   (10+10 Marks)

20. a) Explain the procedure of standardizing the regression model using the

(i) Unit normal scale and (ii) Unit length scale                 (5+5 Marks)

1. b) Explain probit and complementary log-log model (10 Marks)
2. Explain the following methods for scaling the residuals.

(i)  Standardized residuals

(ii) Studentized residuals

(iii) Press residuals

(iv) R Student residuals                                  (5 Marks each)

22. a) Derive the procedure for testing the hypothesis that all of the regression slopes are zero.

(10 Marks)

1. b) Derive the least square estimates of the parameters of a simple regression model.(10 Marks)

 

 

Go To Main page