Alison Gibbs' STA 302 / 1001 Page -- Assignments

STA 302H1F / 1001HF -- Methods of Data Analysis I -- Fall 2011

Assignment 1
Data file (plain text)
Solutions to Assignment 1 (Typo in the solution to question 4 corrected October 13 at 17:45)
Feedback from one of the TAs on common errors made in Assignment 1 Note that in his comment on question 6, he's using the notation of the old textbook where epsilon is what we're calling e (the random error term in the model) and e is what we're calling e-hat (the residual).

Assignment 3
Data file (plain text)
Note regarding the plots in question 1:

The code given on the assignment sheet isn't quite right. Here's what you need. (maxpoints=none) has been added after plots.
ods graphics on;
proc corr plots(maxpoints=none)=matrix(nvar=all);
var bodyfat age weight height adiposity neck chest abdomen hip thigh knee ankle bicep forearm wrist;
run;
ods graphics off;
The plot will be stored in MatrixPlot.png. It will be stored in your home directory, as long as you've selected "Create Listing" in SAS in Tools > Options > Preferences > Results.
SAS will create a scatterplot matrix that has at most 10 variables. Since you have more than that, you are going to need to split the explanatory variables into (at least) 2 groups for the pairwise scatterplots (but you'll want to run proc corr with all of the variables at least once so that you have all of the pairwise correlations). Also, for the plots, include the response variable each time.
This new code works for the version of SAS in the Cquest lab (which is version 9.3). If you're using version 9.2, use the code as originally given on the assignment sheeet (without the maxpoints=none).