Try posting this in the Statistical Procedures forum, you are more likely to get a response. ... View more

Let me see if I understand. You want the denominator to be the number of respondants for the question, but you have eliminated all of the 'N' values. If you had Y and N in your data the ROWPCTN would work. [pre]*Build data; data survey; do ID = 1 to 4; do question = 1 to 5; if ranuni(999999)>.7 then resp='Y'; else resp='N'; if ranuni(99999)>.3 then output; end; end; run; proc freq data=survey; table question*resp; run; proc tabulate data=survey; class question resp; table question,resp*(n rowpctn); run;[/pre] ... View more

For the %DO loop your syntax is off a bit. Try [pre]%do k = 3 %to 5; [/pre] The syntax for the data set name e.g. [pre] work.&data1.winter [/pre] should be fine provided that &DATA1 resolves tothe first part of your data set name and does not contain trailing blanks. For instance if &DATA1 contains ALL your data set becomes: [pre] work.ALLwinter [/pre] but who wants to work all winter anyway. ... View more

One way is to first sparse the data. You can do this by creating a template of all combinations of state and date (I hate using the variable name date for non-sas dates so I have used YRQTR here). [pre]* Build template of data; data allvals(keep=yrqtr state); do state = 'CA', 'PA'; do year = 1976 to 1977 by 1; do qtr = 1 to 4; yrqtr = cats(put(year,4.),':',put(qtr,1.)); output allvals; end; end; end; run;[/pre] The resulting dataset WORK.ALLVARS can now be merged or joined onto your original data. Now working with the sparsed data (like with the LAG function) will be a lot easier. ... View more

The special variable _INFILE_ only contains the information from the input buffer. Although the information from the input buffer is written to the Program Data Vector, PDV, _INFILE_ does not receive any information from the program data vector. Modified variable values are contained in the PDV not in _INFILE_ as is shown here: [pre}data a; input x z$; put _infile_; put x=; x = x+1; put _infile_; put x=; datalines; 1 a 5 b run;[/pre] You need to write the values directly from the PDV not the input buffer. ... View more

Well it depends on how you are writing the values to the graph. I am unfamiliar with the SUBPERCENT option in GCHART. Could you show me your GCHART step? ... View more

You can approach this more than one way. Using the INPUT statement you will probably want to be familiar with the trailing @ to read part of an input line while holding onto the line itself. Also you may want to use the # which allows you to read multiple input lines. The key is to read a portion of the input line, hold it with a trailing @ and then use logic to decide how to proceed with that specific line and subsequent lines for that observation. ... View more

A first pass would be to use PROC IMPORT for each sheet to create a work data set for that sheet followed by an APPEND (PROC DATASETS). It should be fairly easy to automate the process for all the sheets. ... View more

The index function looks for the second argument exactly not as individual characters. For that you need the INDEXC function. Try: [pre]%let x = a(b); %put %sysfunc(indexc(&x,())); [/pre] Of course the parentheses are a particularly unfortunate character to have to work with. ... View more

I believe that you are not getting an answer because we do not understand the question - I don't in any case. I don't believe that I understand all the constraints. Are there necessarily only two groups (more than two values in the second column)? Do you always match first to second? The fourth column is always 2, but UI and UR are not related to PO in the first group. Can you expand your example to show what the fourth column means? ... View more

You may also see improved quality by importing your file into PPT as a picture rather than using the paste buffer. The paste buffer can do some unfortunate things to graph quality. ... View more

I have an admitted bias against stacked histograms, I prefer GROUP= to SUBGROUP= where the error bars make more sense. That said you should be able to calculate the height of each subgroup and therefore the location to place the bars using annotate. ... View more

Syntax is important. As Robert pointed out you have again dropped the &, which is essential to identify the macro variable &YEAR. Without the & the portion of the expression year=1997 will always be false and the %ABORT will not be executed. Are you sure that %ABORT will do what you want? ... View more

Peter, I do not know of an option to retrieve the owner, but would the creation date in FINFO be helpful? Art ... View more

Look up the WHERE statement/clause/option. WHERE data set option. [pre] proc whatever data=mydat(where=(gender=0)); [/pre] WHERE statement. [pre] proc whatever data=mydat; where gender=0; [/pre] ... View more