Ann Arbor ASA Up and Running Series: SAS Sponsored by the Ann Arbor Chapter of the American Statistical Association and the Department of Statistics of.

Ann Arbor ASAUp and Running Series:

SAS

Sponsored by the Ann Arbor Chapter of the American Statistical Association and the Department of Statistics of the University of Michigan

Contents• Starting SAS• User Interface• Libraries• Syntax• Getting Data into SAS• Examining Data• Manipulating Data• Descriptive Statistics• Graphing Data• Statistics in SAS

Up and Running Series: SAS2

Starting SAS

Start SAS 9.3 (English)


User Interface

Log

Comments, warnings, etc.

Program Editor:

Write and submit commands

Output (not seen)

Explorer/ Results

Up and Running Series: SAS 4

Libraries

• SAS requires the creation of Library folders to save the data– Libraries are accessed through LIBNAME command

• Four Libraries are defined by default, at the start of SAS– Maps– SASHELP: holds help info and sample datasets– SASUSER: holds settings, etc.– WORK: default temporary Library for each session

• All data stored in this folder will be deleted at the end of each SAS session

• It is recommended the creation of permanent files/Libraries


Libraries

• Create a folder called ‘my_files’ on your desktop.

• Run this command in SAS: LIBNAME a "C:\Users\uniquename\Desktop\my_files";

• Refer to datasets in that folder by with the prefix ‘a.datasetname’.

• TIP: Use memorable names for libraries, rather than ‘a’ (e.g., ‘raw’, ‘final’, ‘time1’, etc)


Syntax

• SAS divides commands into two groups– DATA step

• create/alter datasets– PROC (Procedures)

• perform statistical analyses or generate reports.

• Some exceptions to the rule:– DATA step can be used to generate reports– PROC IMPORT creates a data set– PROC SORT alters data sets

(without telling you!)


• PROC IMPORT– Allows the reading of standard file types– Allows the reading of plain text, with user-specified

delimiters (i.e., the characters which separate the data)

– WARNING – SAS changed PROC IMPORT for Excel and Access files, in 64-bit SAS

• DATA step– Allows the reading of non-standard file types, complex

file structures, and unusual delimiters.

Getting data into SAS


DATA step

• SAS syntax can be used to read in raw data files (.txt, .csv files), specifying which variables to read in, which ones are text/numeric, combining multiple rows into one case, etc.

• However, this is a more advanced topic.– Follow up with an Intro class from CSCAR, or by

going through examples from the literature

(e.g., ‘The Little SAS Book’).


Examining Data

• VIEWTABLE Window– Select dataset icon in Explorer

• PROC CONTENTS– Produces a listing of data set information, including

the variables and their properties

• PROC PRINT– Prints a subset of variables or cases to the output

window


VIEWTABLE Window


PROC CONTENTS

• In the Editor window, type:

PROC CONTENTS data=a.class2;run;

• Highlight the syntax• Submit for processing

– Click on icon of ‘running-man’– Right click on selected syntax

Submit Selection


PROC CONTENTS


PROC PRINT

• In the Editor window, type:PROC PRINT data=a.class2;run;

• Submit for processing


PROC PRINT


Manipulating Data

• Usually done within a data step– Match data sets using a shared key variable– Create new variables, or drop/rename existing

variables– Take one or more subsets of the data– Sort the data by specific variable(s).

• Overwrite existing or create new datasets– PROC SORT– Adding/Removing variables– Merging Datasets


PROC SORT

• In the Editor window, type:

PROC SORT data=a.class2 out=a.class2sorted;

by age descending weight height;run;


• WARNING: PROC SORT alters data– Store in a new dataset

out=‘newdatasetname’;


PROC SORT


Adding/Removing variables

• Create new data set, compute new variables, remove unwanted variables

DATA a.class2metric (drop=weight height sex age);

set a.class2;

height_cm=height*2.54;

weight_kg=weight/2.2;

label height_cm=‘Height in CM’

weight_kg=‘Weight in Kilograms’;run;

PROC PRINT data=a.class2metric;

run;• Submit for processing


Adding/Removing variables


Merging Datasets

• Data sets must be sorted by the same key variable(s)

proc sort data=a.class2;

by name;

proc sort data=a.class2metric;

by name;

data classmerged;

merge a.class2 a.class2metric;

by name;run;



Merging Datasets


Merging Datasets


Descriptive Statistics

• PROC FREQ– Produces a table of counts and percentages– For cross-tabulations, statistical tests can also

be performed; e.g., independence testing

• PROC MEANS– Produces descriptive statistics such as mean,

standard deviation, minimum, maximum


PROC FREQ

• In the Editor window, type proc freq data=a.class2;

tables age*sex;run;



PROC FREQ


PROC MEANS

• In the Editor window, type proc means data=a.class2;

var age weight height;

run;



PROC MEANS


Graphing DataPROC GPLOT

• Simple bivariate scatterplot• Separate lines• Multiple variables scatterplot• Options


PROC GPLOT

• Simple bivariate scatterplot:proc gplot data=a.class2;

symbol1 value=dot interpol=rl;plot weight*height;

run;



PROC GPLOT - Log


PROC GPLOT


• To graph separate lines for each level of a categorical variable, type:

proc gplot data=a.class2;

symbol1 value=dot interpol=rl;

plot weight*height = sex;

run;


PROC GPLOT


PROC GPLOT


• Multiple variables on the same graph:proc gplot data=a.class2;

symbol1 value=dot interpol=rl color=blue;

symbol2 value=dot interpol=rl color=red;

plot weight * age;plot2 height * age;

run; quit;


PROC GPLOT


PROC GPLOT


value=___

• Any character enclosed in single quotes

• Special characters– dot– plus sign– star– square– ...and many others

interpol=___

• RL / RQ / RC– linear– quadratic – cubic – regression curves

• JOIN– connects consecutive

points (line graph)• BOX

PROC GPLOT


Statistics in SAS

• PROC CORR– Correlational analyses

• PROC REG– Statistical Regression

• PROC UNIVARIATE– To assess normality of regression residuals


PROC CORR

• Compute bivariate correlation coefficients

proc corr data = a.class2;var age;with height weight;

run;


PROC CORR


PROC REG• Run a regression on merged ‘class’ dataset

– Save residuals and predicted values in an output dataset

– Request residual plotproc reg data=a.classmerged;

model height_cm=age weight / partial; output out=reg_data p=predict r=resid

rstudent=rstudent; plot rstudent. * height_cm;

run;quit;

• Notes – the quit command terminates the regression procedure; otherwise it keeps running; the output data set will be in the work library, since no library was specified.Up and Running Series: SAS 42

PROC REG


PROC REG


PROC REG


PROC REG


PROC UNIVARIATE

• Assess normality of regression residuals stored in the output dataset from PROC REG:

proc univariate data=reg_data;var rstudent;histogram;qqplot / normal (mu=est

sigma=est);run;quit;


PROC UNIVARIATE


PROC UNIVARIATE


PROC UNIVARIATE


QUESTIONS


Winter 2013 Training from CSCARhttp://cscar.research.umich.edu/workshops/

Introduction to SAS® - January 28,30, February 1,4,6,8, 2013

Intermediate Topics in SPSS: Data Management and Macros - February 5,7, 2013

Intermediate Topics in SPSS: Advanced Statistical Models - February 12,14, 2013

Intermediate SAS® - February 25,27, March 1, 2013

Regression Analysis - March 11,13,15, 2013

Applications of Hierarchical Linear Models - March 18,20,22, 2013

Statistical Analysis with R - March 19,21, 2013

Introduction to NVivo - April 3, 2013

Applied Structural Equation Modeling - April 10,11,12, 2013


Further Resources

• The Little SAS Book: A Primer• UCLA site

– software tutorials, classes and lectures on statistical methods – an incredible site! http://www.ats.ucla.edu/stat/

• SAS Documentation: http://support.sas.com/documentation/

Documentation also found in ‘SAS help’ files.


http://www.ats.ucla.edu/stat/

http://www.ats.ucla.edu/stat/

http://support.sas.com/documentation/

54

Other Winter 2013 Workshopsfrom Ann Arbor ASA

R - January 31, 1-3 PM

Angell Hall Computing Classroom B

(also known as MH444-B)

For more information go to: http://community.amstat.org/annarbor/home

Up and Running Series: SAS

http://community.amstat.org/annarbor/home

PLACE Starbucks State & Liberty, lower level

TIME 6:00pm – 6:45pm,

DATE TOPIC

24-JAN Business Meeting 1 -APR Business Meeting and Election of Officers

For more information go to: http://community.amstat.org/annarbor/home

Chapter Meetings open to all


http://community.amstat.org/annarbor/home

Ann Arbor ASA Up and Running Series: SAS Sponsored by the Ann Arbor Chapter of the American Statistical Association and the Department of Statistics of.

Documents