EXST7034 – Regression Techniques Fall 2003 HOMEWORK ASSIGNMENT 6 Assigned: November 3, 2005 Due: November 10, 2005 20 Points – 1 point each day late All data is in "http://www.stat.lsu.edu/EXSTWeb/statlab/datasets/KNNLdata/". Complete the following questions from your textbook. There are two problems. The first problem is analysis of covariance example based on the "Copier maintenance" example (8.15 & 8.19). The original datasets are in CH01PR20.txt and CH08PR15.txt. As a special treat I have prepared a csv dataset called CH08PR15.csv! The dataset 8.15 has variables Y and X1, but not X2. Dataset 8.19 has X1, but not Y or X1. You may fix this any way you wish. If you prefer you may use CH08PR15.csv as an external file. This external file has all 3 variables separated by commas. To use this dataset you must tell SAS that the separator is a comma. The first line has the values “minutes serviced model”, so you will need to start reading data on the second line. My input segment is given below. filename copier 'C:\Geaghan\Current\EXST7034\Fall2005\Datasets (KNNL)\CH01PR20.csv'; ODS HTML style=minimal rs=none body='C:\Geaghan\Current\EXST7034\Fall2005\SAS\CH01PR20A6.html' ; DATA copier; infile copier missover DSD dlm="," firstobs=2; INPUT minutes machines model; The second problem is a variable selection example based on the patient satisfaction example (9.17). The dataset is in CH06PR15. A) 8.15 a - e Note that for this first part no interaction is fitted. The first model is the textbooks 8.33 (011 22iiiiYXXβββ ε=+ + +) where X2 is an indicator variable. 1) Fit the model and complete questions a, b and c. Answer the question in d. 2) Plot the residuals on X1X2 (interaction) as suggested in the book. Is there an indication of an interaction? Try a second plot using the statement below. Note I placed the residuals, which I called "E", in a data set that I called "RESIDS". PROC PLOT DATA=RESIDS; PLOT E*X1=X2 / VREF=0; RUN; B) 8.19 a & b Note that for this second part an interaction is fitted. 1) Complete both parts of problem 8.19. 2) Can you figure out how to estimate both slopes and both intercepts with standard errors in a single model with pooled variance? C) 9.17 a, d & e Do only parts a, d and e with the following alterations. a) Use the stepwise selection options and a value of “SLENTRY = 0.05” and “SLSTAY = 0.05”. b and c) You do not need to answer these question. d) Use the backward selection options and a value of “SLSTAY = 0.05”. e) Compare the two techniques used in parts a and d above.EXST7034 – Regression Techniques Fall
View Full Document
Unlocking...