topic forest plot in SAS Studio

forest plot

RamJ — Thu, 03 Mar 2016 21:00:46 GMT

The following program is printng all text and numbers in unbolded font. How can I have it print everything in bold. Also, the bars are printed on the top of the symbol. Can I have the symbol printed on the top of the bar? Also, is there a way to have a "blank" line inserted in-between certain bars? Here is the data and program I am using?

data x;
input sname $15. AGM LL UL ns2 na2;
format AGM LL UL f5.1;
datalines;
SCOT: Males 233.1 182.9 297.1 12 12
SCOT: Females 241.3 189.1 307.9 11 11
SCOT: NHW 218.4 170.8 279.4 10 10
SCOT: NHB 294.1 232.8 371.5 9 9
SCOT: HISP 217.9 167.0 284.3 8 8
SCOT: OTH 226.1 174.2 293.4 7 7
NNAL: Males 258.0 162.1 410.7 6 6
NNAL: Females 360.5 226.7 573.3 5 5
NNAL: NHW 334.9 214.9 522.0 4 4
NNAL: NHB 258.5 157.0 425.5 3 3
NNAL: HISP 324.1 202.9 517.8 2 2
NNAL: OTH 308.3 189.9 500.6 1 1
;
run;
title 'Adjusted Geometric Means with 95% Confidence Intervals';

proc sgplot data=x noautolegend nocycleattrs;
refline 150 600 / axis=x;
scatter y=ns2 x=agm / markerattrs=graphdata2
(symbol=circlefilled size=15 color=black);
highlow y=ns2 low=ll high=ul / type=bar intervalbarwidth=.05in
nooutline fillattrs=graphdata1 (color=Green);
yaxistable sname / y=ns2 position=left location=outside
nolabel valuejustify=left pad=(right=0px) labelattrs=(weight=bold);
yaxistable agm LL UL / y=nS2 position=right location=inside
labelhalign=center labelattrs=(weight=bold)
pad=(left=20px right=10px);
yaxis display=none offsetmin=0.1 offsetmax=0.05 values=(1 to 12) ;
run;

Ram Jain

Re: forest plot

hollandnumerics — Thu, 03 Mar 2016 22:12:59 GMT

Ram,

Sometimes in life "size matters", but in ODS Graphics "order matters"!

The following code has been updated to:

split sname into 2 parts, so that your input statement reads the data you expect
reorder scatter after highlow, so the scatter is plotted after highlow (and on top of it!)

data x;
    length sname1 sname2 $7. sname $15.;
    input sname1 sname2 AGM LL UL ns2 na2;
    format AGM LL UL f5.1;
    sname = strip(sname1) || ' ' || strip(sname2);
    datalines;
SCOT: Males 233.1 182.9 297.1 12 12
SCOT: Females 241.3 189.1 307.9 11 11
SCOT: NHW 218.4 170.8 279.4 10 10
SCOT: NHB 294.1 232.8 371.5 9 9
SCOT: HISP 217.9 167.0 284.3 8 8
SCOT: OTH 226.1 174.2 293.4 7 7
NNAL: Males 258.0 162.1 410.7 6 6
NNAL: Females 360.5 226.7 573.3 5 5
NNAL: NHW 334.9 214.9 522.0 4 4
NNAL: NHB 258.5 157.0 425.5 3 3
NNAL: HISP 324.1 202.9 517.8 2 2
NNAL: OTH 308.3 189.9 500.6 1 1
;
run;

title 'Adjusted Geometric Means with 95% Confidence Intervals';

proc sgplot data=x noautolegend nocycleattrs;
    refline 150 600 / axis=x;
    highlow y=ns2 low=ll high=ul / type=bar intervalbarwidth=.05in
        nooutline fillattrs=graphdata1 (color=Green);
    scatter y=ns2 x=agm / markerattrs=graphdata2
        (symbol=circlefilled size=15 color=black);
    yaxistable sname / y=ns2 position=left location=outside
        nolabel valuejustify=left pad=(right=0px) labelattrs=(weight=bold);
    yaxistable agm LL UL / y=nS2 position=right location=inside
        labelhalign=center labelattrs=(weight=bold)
        pad=(left=20px right=10px);
    yaxis display=none offsetmin=0.1 offsetmax=0.05 values=(1 to 12);
run;

Enjoy!................Phil

PS. Look for "Philip R Holland" on Amazon - you may enjoy that too!

Re: forest plot

RamJ — Thu, 03 Mar 2016 23:41:34 GMT

Thank you so much but it still prints text on the Y-axis like SCOT: Males
and all numbers like 226.1 etc in unbolded font. Is there a way to print
these also in bold? Also, is there a way to insert blank lines, for
example, after SCOT: OTH.

##- Please type your reply above this line. Simple formatting, no
attachments. -##

Re: forest plot

hollandnumerics — Fri, 04 Mar 2016 10:28:52 GMT

Ram,

A few amendments marked in red to match your specification:

data x;
    infile datalines truncover;
    length sname1 sname2 $7. sname $15.;
    input sname1 sname2 AGM LL UL ns2 na2;
    format AGM LL UL f5.1;
    if sname1 ne '-' then sname = strip(sname1) || ' ' || strip(sname2);
    datalines;
SCOT: Males 233.1 182.9 297.1 13 13
SCOT: Females 241.3 189.1 307.9 12 12
SCOT: NHW 218.4 170.8 279.4 11 11
SCOT: NHB 294.1 232.8 371.5 10 10
SCOT: HISP 217.9 167.0 284.3 9 9
SCOT: OTH 226.1 174.2 293.4 8 8
- - . . . 7 7
NNAL: Males 258.0 162.1 410.7 6 6
NNAL: Females 360.5 226.7 573.3 5 5
NNAL: NHW 334.9 214.9 522.0 4 4
NNAL: NHB 258.5 157.0 425.5 3 3
NNAL: HISP 324.1 202.9 517.8 2 2
NNAL: OTH 308.3 189.9 500.6 1 1
;
run;

options missing = ' ';

title 'Adjusted Geometric Means with 95% Confidence Intervals';

proc sgplot data=x noautolegend nocycleattrs;
    refline 150 600 / axis=x;
    highlow y=ns2 low=ll high=ul / type=bar intervalbarwidth=.05in
        nooutline fillattrs=graphdata1 (color=Green);
    scatter y=ns2 x=agm / markerattrs=graphdata2
        (symbol=circlefilled size=15 color=black);
    yaxistable sname / y=ns2 position=left location=outside
        nolabel valuejustify=left pad=(right=0px) labelattrs=(weight=bold)
        valueattrs = (weight=bold)
        ;
    yaxistable agm LL UL / y=nS2 position=right location=inside
        labelhalign=center labelattrs=(weight=bold)
        valueattrs = (weight=bold)
        pad=(left=20px right=10px);
    yaxis display=none offsetmin=0.1 offsetmax=0.05 values=(1 to 13);
run;

Does that work for you now?................Phil

PS. BTW you don't use na2 anywhere in this program!