About AncaTilea

AncaTilea · ‎11-28-2012

Hello SAS Experts, Masters and Gurus! I am trying to create a forestplot-like graph, using PROC TEMPLATE. Most things work as I expect to work. However, I have 2 questions: (1) I would like to ident the "subgroup" labels on the y-axis, like this: Gender Male Female ... (2) I would like the lines (mean +/- s.e.) to alternate between solid and dashed (so, gender has lines black and solid, race has lines gray and dashed...) My code is below and the sample data is attached: /*hoping to get the leading-blanks to stay in the format*/ data _null_; call symput('blk','A0'X); run; proc format; value $ lab 'Male' = "&blk.&blk. Male" 'Female' = "&blk.&blk. Female" 'Black' = "&blk.&blk. Black" 'White' = "&blk.&blk. Non-Black" ' 43 years' = "&blk.&blk. 43 years" ' 66 years' = "&blk.&blk. 66 years" ' 82 years' = "&blk.&blk. 82 years" ; quit; proc template; define statgraph sgdesign; dynamic _ESTIMATE _LABEL _GRP_VAR _ESTIMATE2 _UPPER _LOWER; begingraph / designwidth=1408 designheight=807; DiscreteAttrMap name="__ATTRMAP__"; Value "1" / markerattrs=( color=black symbol=circle) lineattrs=( color=black pattern=2); Value "2" / markerattrs=( color=grayaa symbol=plus) lineattrs=( color=grayaa pattern=3); Value "3" / markerattrs=( color=black symbol=circle) lineattrs=( color=black pattern=2); EndDiscreteAttrMap; DiscreteAttrVar attrvar=my_grp var=_grp_var attrmap="__ATTRMAP__"; layout lattice / rowdatarange=data columndatarange=data rowgutter=10 columngutter=10; layout overlay / xaxisopts=( display=(TICKS TICKVALUES LINE LABEL ) griddisplay=off label=('Mean (95% Confidence Interval) of eGFR Slope') labelattrs=(size=14 style=NORMAL weight=BOLD ) tickvalueattrs=(size=14 )) yaxisopts=( reverse=true display=(TICKS TICKVALUES LINE ) griddisplay=off labelattrs=GraphTitleText(size=14 ) tickvalueattrs=(size=14 ) gridattrs=(color=CX848284 pattern=2 thickness=1 )); scatterplot x=_ESTIMATE y=_LABEL / group=my_grp datalabel=_ESTIMATE2 datalabelposition=TOP DATALABELATTRS = (size = 14) xerrorupper=_UPPER xerrorlower=_LOWER name='scatter' markerattrs=(symbol=CIRCLEFILLED size=15 weight=bold ) errorbarattrs=(thickness=2 ); referenceline x = 0 / lineattrs = (thickness=3 pattern=2 color=CX848284); endlayout; endlayout; endgraph; end; run; proc sgrender data = have template = sgdesign; dynamic _ESTIMATE = "ESTIMATE" _LABEL = "LABEL" _GRP_VAR = "'GRP_VAR'n" _ESTIMATE2 = "ESTIMATE" _UPPER = "UPPER" _LOWER = "LOWER"; run; Thank you!!!!!

AncaTilea · ‎11-26-2012

ok,all, I think here is a simpler code using the not alpha numeric function. /*NOTALNUM Purpose: To determine the position of the first character in a string that is not an alphanumeric (any upper- or lowercase letter or a number). If none is found, the function returns a 0. With the use of an optional parameter, this function can begin searching at any position in the string and can also search from right to left, if desired. */ data have; input x $ ; cards; 1a 1234g 3h j6 4 5 6 . .. . ... . 4 4 ; data want; length value $10.; set have; temp = NOTALNUM(x); if temp = 1 then value = x; else if temp = 2 then value = "Number"; else value = "Non-Number"; run;

AncaTilea · ‎11-26-2012

Agreed. BUT My understanding was that silasskovsbo wanted to focus on the missing data (basically wanted to find out how many ways the missing data was coded in). If that's not indeed correct then the PROC FORMAT needs to be modified accordingly. Cheers!

AncaTilea · ‎11-26-2012

Is this what you need? data have; input x $ ; cards; a 4 5 6 . .. . ... . 4 4 ; proc format; value $ fmt 0-99999 = "Number" ; data want; set have; format x $fmt.; run; proc freq data = want;table x;run; So in this way, if you do a frequency you get what you want. (given that I understood what you need) Let me know if you have any questions. Good luck!

AncaTilea · ‎11-24-2012

Hi. If you want to "linearly" impute the data then you can do a proc reg on your data, save the linear equation, then calculate the missing values. Something like this: *using the data you have, I create "order" variable which is really just 1..n; data have; set have; order = _n_; run; *run a linear regression to get the parameters for the line; proc reg data = have; model value = order; ods output parameterEstimates = est(keep = variable estimate); run;quit; *save the intercept and the beta; data _null_; set est; if variable = "Intercept" then do; call symput ("int", estimate); end; if variable = "order" then do; call symput ("beta", estimate); end; run; data want; set have; if value = . then value = &int. + order * &beta.; run; Good luck!

AncaTilea · ‎11-16-2012

from quickly scanning the internet, it seems that there is no automatic way to calculate it. 😐

AncaTilea · ‎11-16-2012

Here is a helpful link: http://listserv.uga.edu/cgi-bin/wa?A2=ind0711a&L=sas-l&P=8219 data _null_; gmt = '20:05:09't; est = gmt - (4*60*60); pst = gmt - (7*60*60); format _all_ time.; put _all_; run; and here is another discussion:

AncaTilea · ‎11-16-2012

Maybe you can provide some sample code, and/or the output associated with it. 🙂

AncaTilea · ‎11-12-2012

Hi. There were a couple of errors in your code. As said before the %let df_ind = 4 confuses SAS, because in your macro you start with a DO-LOOP that resolves to; do 4 = 4 to 15; Nonsense. So I modified your code (I used sashelp.class as data set since I didn't have your data) First, in the DO-LOOP your df_ind is not a macro variable for the macro PlotFit. I added a PROC SORT so I can make some meaning in what I am plotting; I added a "data =" in your PROC GPLOT; (Please note that you need to change the bolded text in the code below) %MACRO PlotFit; %do df_ind=2 %to 3; PROC GAM data= sashelp.class; model weight=spline(age,df=&df_ind.)/DIST=GAMMA; output out=Fit&df_ind._M p; run; proc sort data = Fit&df_ind._M;by p_weight; title "Fit with df=&df_ind."; proc gplot data = Fit&df_ind._m; plot P_weight*age weight*age/overlay; symbol interpol=join l=1 w=2; run;quit; %end; %MEND PlotFit; %PlotFit; Best of luck! Anca.

AncaTilea · ‎11-11-2012

Thank you PG for clarifying. I modified your code to see what's happening; data want; set have; by var2 notsorted; *if not (first.var2 and last.var2); if first.var2 = 1 and last.var2 = 1 then to_delete = 1; run; so I can see clearly what gets deleted. Thanks!

AncaTilea · ‎11-11-2012

I don't understand how did you summarize the data set that you say you want. What happened to the values 1,3,4 and 5? When you say you want all the same values in the variable, you mean var1 or var2 in your data set? Please give a bit more detail. Thanks!

AncaTilea · ‎11-06-2012

Sure, you could display only the value for the subgroup. One dirty way to do that (given that you have exactly 12 months of data in each year) would be to assign blank values to the tick marks, and the 2008, 2009 or 2010 value for some of the tick marks. See below: goptions reset = all; axis1 v = ("" "" "" "" "" "2008" "" "" "" "" "" "" "" "" "" "" "" "2009" "" "" "" "" "" "" "" "" "" "" "" "2010" "" "" "" "" "" ""); proc gbarline data=temp; bar date1 / discrete sumvar=y1v subgroup=year maxis = axis1; plot / sumvar=y2v; run;quit;

AncaTilea · ‎11-06-2012

You could use a subgroup = your_variable for the bar plot.; but first you need to sort your data by that group variable. Here is an example using the SASHELP.CLASS data set. proc sort data= sashelp.class out = temp;by sex; proc gbarline data = temp; bar name / discrete sumvar = age subgroup = sex; plot / sumvar = weight ; run; quit; Best of luck. Anca.

AncaTilea · ‎11-06-2012

Using the SASHELP.CLASS data set as an example. There are two genders (M and F) and age and weight as continuous. In first plot, i want the age * weight by gender, so I use the symbol1 i = join and repeat = 2(for my 2 genders), so for you repeat = 369 I assume is for the total number of IDs. Then I do the plot2, for the regression line (overall) between age and weight, using the symbol2 statement with i = r. So, the only difference between your code and mine is where I have the symbol statement (immediately after the corresponding plot statement).(which should not matter) goptions reset=all; proc sort data = sashelp.class out = temp;by weight; proc gplot data = temp; plot age*weight = sex; symbol1 v = dot repeat = 2 i = join color = blue width = 3; plot2 age*weight; symbol2 v = none i = r color = green width = 2; run;quit; Would you be able to provide a sample data?

AncaTilea · ‎11-06-2012

ods graphics on; proc corr data= sashelp.class nomiss plots = scatter (nvar=2 alpha=.20 .30); var age height; run; ods graphics off; What version of SAS do you use?

Online Status	Offline
Date Last Visited	‎11-18-2024 07:33 PM

Re: Crosstab Dates

Re: why repeated observations??

Re: why repeated observations??

Re: why repeated observations??

Re: Add rivers to US map

Re: Add rivers to US map

Add rivers to US map

Re: # of months between two date vars

Re: ODS Tagsets ExcelXP Creating Multiple Workbooks

Re: Histogram with different fill colors beyond a specified point

Re: Add rivers to US map

Re: Spliting a string into variable

Re: Spliting a string into variable

Re: Histogram with different fill colors beyond a specified point

Re: Anagrams Within Words

Re: Default "view" preferences at startup of SAS

Re: Turn off title in BoxPlots

Re: How to use "labels" (Excel column headers) instead of SAS "names" ...

Re: Residuals in logistic regression

Re: Adding a Variance column in a table

ForestPlot with PROC TEMPLATE: indentation issue and group attributes

Re: proc format

Re: proc format

Re: proc format

Re: linear interpolation of missing data

Re: Is there a way to convert time from EST to GMT?

Re: Is there a way to convert time from EST to GMT?

Re: pdf header issue.

Re: Plots loop in macro

Re: Grouping same values together

Re: Grouping same values together

Re: gbarline bars

Re: gbarline bars

Re: Proc Gplot ignoring symbol2

Re: Need help to solve an ODS Error