SAS tour

1. It is important to see how to get data "into" SAS. So look at the
easiest way first: put the data IN the sas program.

I put a test program called "babyData.sas" in my SAS working example
code directory.  Download that, open it in the SAS program editor, add
some more lines of data just for fun, and then run the PROC MEANS and
PROC GPLOT statements.


When you run that, look at the LOG window for error messages. Look at
the Output window for, hmm, I forget what's in there. A separate
graphic window should come up too. See?

Now, in SAS for the PC, there is an interactive thing on the left
called "Explorer". It shows data libraries, and if you open the
"libraries" button with a double click, you should see some
folders. The "Work" folder is the place where it is storing your
temporary datasets in sas format. Open work, doubleclick on your
dataset and see what you see! If you don't see something, make a note!

Once the gplot output shows, try to save a picture of the image as a
file. Choose File/export as image. For god's sake, don't choose bmp
format. Choose gif for a picture format and/or choose epsi or ps for a
format that many document preparation programs can incorporate.

If you save a gif or jpeg format picture, use a graphic editor like
irfanview to open up the picture, cut or color it, do whatever you
want. Just convince yourself you did create a picture.

Gplot output can be customized in a million ways, to control the
plotting character, the axis labels, and so forth.  I'm not an expert,
but if I look in the SAS manuals online on lark, I can find examples
that do what I like. I urge you to look there!

In the editor, change PROC GPLOT
statement and PROC PLOT, and look in the output window. 
Do you see a difference?

2. Next consider how to create new variables and do recodes.  In my
example code directory, I have 2 pretty big fully functional
programs. One is called "crime.sas" (there's a readme for it) and
there is "bank.sas". You can download/run those if you want. But study
the data steps! Look at the recode lines, which come BEFORE the cards
statement.  I don't know why, they just do!

So take your copy of babyData1 and create some new variables, make a
dummy variable, create a new variable that uses a mathematical
function of one or more x's.  You can do things like

myNewV1 = 3*x1;
or
myNewV2 = log(x2);

or

if (x1 > 3) then myNewV3 = 0;
else myNewV3=1;

So just create some new variables, recode some stuff.

If you want to set something as missing, set a variable equal to a
period. As in (I'm using the word "eq" for equals)

if (x2 EQ 99) then x2=. ;

That makes x2 missing when the original value was 99.

So put in some phony number in your data, then take it out that way,
and compare the plots before and after.

3. If your data is a big file, you don't want it in the middle of your
program. 

THere are 2 options. Use the INFILE statement to grab the data from
a file. That's described in the Little SAS Guide. I've got an example
of its usage in the baseball directory. The key thing is to see that I
give the file name in single quotes. That works if the data file is in
the same directory as your sas file. As the book explains, it is
necessary to give the full path to the data file, you know, something
like 'd:\myStuff\ps707\crimedata\' or whatever.

Another option is this cool thing Dave Brichoux found out last year.
If you have the sas statements that create data step and it works
fine, you can save that stuff in a separate file and then use an
include statement, that grabs that whole file and copies it all into
your current program. For example, if you look at this example code:

http://lark.cc.ku.edu/~pauljohn/SASClass/ExampleCode/loess1.sas

You should see this one:

%include 'bank.sas';

So take either route to separate your data in babyData1.sas from your
sas code.


4. Learn how to get a frequency distribution out of SAS.  I usually
just use PROC FREQ.  They have a manual on that. :)

5. In SAS, PROC FREQ is also used to create crosstabulation tables.

6. If you are interested, there is proc chart for making nice looking
histograms and bar charts. Again, there is an online manual on lark. 
Spend a little time digging around in those docs.


At the end of this, I want you to give me 
a) a scatterplot, 
b) a printed out copy of a sas program that does some recodes and
creates some new variables,
c)  the sas program that you use in part 3.
d) your choice of a crosstabulation, a frequency distribution, or
a chart.

Special things to note:
"Clear All" under Edit.
See a way to save your sas programs?
See a way to save a data set in the special sas format? See a way to
reload it? (this is a hard one, look for the keyword "libname").