SAS Chapter 8 - University of South Carolina

Download Report

Transcript SAS Chapter 8 - University of South Carolina

Chapter 8: ODS Graphics


ODS graphics were not available prior to
SAS 9.2
They have been implemented across a
wide range of procedures (reminiscent of
JMP)
© Fall
2011 John Grego and the University of South Carolina
1
Chapter 8: ODS Graphics




Graphs can be routed to different devices
(default is LISTING)
For default output, graphs are available in
the Results window as separate objects
The default graph type is .png; it can be
modified to .jpg, etc
Some style choices are readily edited
© Fall
2011 John Grego and the University of South Carolina
2
PROC SGPLOT


Omnibus procedure to create ODS
graphics
We will review
− Bar charts
− Histograms and box plots
− Scatterplots
− Time series plots
− Fitted curves
3
PROC SGPLOT-Bar Charts



Creates vertical or
horizontal bar charts
Groups can be added
Bars can actually be
summary statistics for
continuous variables
PROC SGPLOT;
VBAR varname/
GROUP=catname;
VBAR varname/
RESPONSE=varname
STAT=stattype;
LABEL
varname=‘varlabel;
4
PROC SGPLOT-Histograms


SCALE subcommand
graphs percents, counts,
or proportions on the
vertical axis
DENSITY overlays a
normal curve (default) or
kernel density estimate
PROC SGPLOT;
HISTOGRAM varname;
DENSITY varname;
DENSITY
varname/TYPE=kernel
;
5
PROC SGPLOT-Boxplots


VBOX specifies graphing
variables
CATEGORY generates
side-by-side plots
PROC SGPLOT;
VBOX varname/
CATEGORY=catvarname;
6
PROC SGPLOT-Scatter Plots




Alternative to PROC
GPLOT
Multiple SCATTER
statements produce
overlaid plots
LABEL improves the
legend for overlaid plots
GROUP creates different
plotting symbols
PROC SGPLOT;
SCATTER X=hname
Y=vname;
GROUP=groupname;
XAXIS LABEL=‘xlabel’
VALUES=(low TO high
BY increment);
7
PROC SGPLOT-Time Series Plots



Data points are
connected by lines by
default
The horizontal axis will be
time, date, etc
LABEL command for
dates follows naming
conventions we have
seen before
PROC SGPLOT;
SERIES X=timeindex
Y=vname;
GROUP=groupname;
XAXIS LABEL=‘xlabel’
VALUES=(‘ddmmmyy’d
TO ‘ddmmmyy’d BY
increment);
8
PROC SGPLOT-Fitted Curves



A scatterplot is
automatically generated
for REG, LOESS, or
PBSPLINE
CLM adds confidence
bounds
NOMARKER suppresses
the oft-redundnat printing
of the scatter plot
PROC SGPLOT;
REG X=hname Y=vname;
LOESS X=hname
Y=vname;
PBSPLINE X=hname
Y=vname;
9
PROC SGPANEL

Panel graphs are an
increasingly preferred
alternative to traditional
methods for multi-graph
displays.
PROC SGPANEL ;
PANELBY variablelist;
SCATTER X=hname
Y=vname;
SERIES X=index
Y=vname;
Etc.
10