ST500       January 3, 1996    Session 2 Class will be held from 10-noon
in C141, 1st Floor of Clark The CTSS Lab will also be open 9-1 weekdays.
Assignments are due beginning of 2nd class period

DATE            TOPIC

Tue 2  Intro/mail/pico/ftp + telnet/unix

Wed 3  Unix/ftp/vi + pico/.login file/printing/sas commands
                 assign 1(a)-sas
Thu 4  Review/spss commands
                assign 1 (b)-spss
Fri 5  minitab commands/pc stat pgm info
                 assign 1(c)-minitab

Mon 8  Regression: sas/spss/minitab
                 assign 2-regression
Tue 9  Analysis of Variance:sas/spss/minitab
                 assign 3-anova
Wed 10              Sas - Interactive Display Manager
                 assign 4-land appraisal
Thu 11              anov + reg options/diagnostics

Fri 12 Xwindows:lnsight(sas)/bmdp/xbmdp

Grading: assignments 1-4 @ 20% each =80%
             attendance =20%



                          Second session

_Unix: man command>f1
       A Practical Guide to the Unix System, Mark G. Sobell  $27.95 Unix
       in a Nutshell, O'Reilly & Assoc  $19.50
_Get on Vector mailing list at ACNS and setup grad student account _ftp
: use either c:\temp to read and write files on the lab pcs _vi: full
screen editor + pico

_printing from lamar at studlO/msite  (lpr -PstudIO or lpr -Pmsite)
_printing from lamar in C141 - need to use ftp

_sas commands: syntax + examples

_assign 1 (a) - sas (note that this is due 2 class days later: Jan 5 at
the beginning of class)ST500     Assiqnment 1 (a) - sas

Assignments 1(b) and 1(c) will be similar to this exercise, but use
spss and minitab, respectively.  Whenever you begin data analysis with a
computer package, you must first check your data for invalid values or
'outliers', make sure that the data fields are being read correctly,
verify that missing data, if any, is treated appropriately, and decide
if transformations of the response variables might be appropriate. The
tasks below are suggested as some ways in which to do that checking.

The following tasks should be done with SAS using the assign1.dat data
file. To get a copy in your home directory: cp ../data/assign1.dat
assign1.dat. You need to email (or print) the (1)control statements
(2) xxx.log file by beginning of class on Jan 5.

1. Read assign1.dat into SAS then list it. (NOTE: the file "../assign/a1a"
should be a tremendous help!)  2. Do a frequency distribution of all data
fields. (PROC FREQ) (By now you should have noticed a negative value
for weight. It should be treated as missing for further analyses. IF
.. THEN .. =.;) ( The 'small' value for weight should be 84.0, use vi
or pico to fix the data file.) 3. Compute means/minimums/maximums of
height,weight and age for m and f (sex) separately. (PROC MEANS; CLASS
..;) 4. Create two SAS data files: one with m and one with f. (DATA ..;set
..;IF ..;) 5. Recombine the two files into one file and print.(DATA ..;
set ....;) 6. Sort the file by name within sex, and print. (PROC SORT:
BY ....; 7. Plot weight against height and age, using two symbols on the
plot to indicate sex. (PROC PLOT; PLOT.. *.. =..; 8. compute LOG10 of
weight (Iwt), then plot Iwt against age with symbols for sex as in 7.
(Iwt=log10(weight) ;) 9. Use PROC UNIVARIATE to get stem + Leaf plot,
boxplot and normal probability plot of weight, height, and Iwt. (PROC
UNIVARIATE PLOT; VAR ..;)

Now without using the assign1.dat data, we will look at several data
distributions.  Remember if you see histograms similar to those for C
and D, the data can be transformed to 'normal' with a square-root or
log transformation. Other common transformations are the inverse (1/y)
and arcsin-sqrt(p).

10. Generate 200 random normal observations, A. {DATA temp; DO i=1 TO
200; A=NORMAL(1234567); output;end;} Then compute B=A*.2+1  {Normal
(1,.04)}.  Then compute C=B*B.  Then compute D=10**(B).  Finally, run
PROC UNIVARIATE as in 9) for variables A,B,C and D.