I had to match two files based on 4 different keys from most strict to loose to add a cost variable. I did this in steps using proc sql and created 4 different tables. Then I appended the tables together with the non-matches so that I had the table I started with but with match_flag1, match_flag2, match_flag3, match_flag4.  I only want to count the first occurrence of the data that I retrieved from the other table based on the hierarchy of best match.  I can't figure out how to do this and keep the original data. I feel like Hash Add() could do this if I output to 4 different tables and add them together again but I need to keep all the data in place so that duplicated reasoning can be maintained for each row. In other words I cannot output a separate table. I just want to update the value if it is the first occurrence of the match.  For example, if match_flag3=1 for one row of the key and match_flag4=1 for a diffferent row, I only want to keep the value in the row where match_flag3 =1 for that key. I tried doing this with first. and last. logic but becuase there are 4 different types of matches I could not get the right sort and logic.    The following logic almost gets me there but there could still be duplicates keys when match_flag# variables =1 so I need to still identify the first occurrence of the key  within the following logic.    data test; if _N_ = 1 then do; declare hash m1 (dataset: "Have"); m1.defineKey("id", "dos", "diag", "provider", "lob"); m1.defineData("match_flag1"); m1.defineDone();   declare hash m2 (dataset: "Have"); m2.defineKey("id", "dos", "diag", "provider"); m2.defineData("match_flag2"); m2.defineDone();   declare hash m3 (dataset: "Have"); m3.defineKey("id", "dos", "diag", "lob"); m3.defineData("match_flag3"); m3.defineDone();   declare hash m4 (dataset: "Have"); m4.defineKey("id", "dos", "diag"); m4.defineData("match_flag4"); m4.defineDone(); end;    set Have (drop=match_flag1 match_flag2 match_flag3 match_flag4 );   length match_flag1 8. match_flag2 8. match_flag3 8. match_flag4 8.; if m1.find() = 0 then do; value_revised =value; match_keep=1; end; else if m2.find()=0 then do; value_revised = value; match_keep=1; end; else if m3.find()=0 then do; value_revised=value; match_keep=1; end; else if m4.find()=0 then do; value_revised =value; match_keep=1; end; else do; match_keep= 0; value_revised = 0; end; ; run;   ... View more

hello, I am trying to show two lines for each group in the group= statement of an sgplot.  One line would be the actual value and the other would be the 12 month moving average. I want the actual to be a solid line and the moving average to be a dashed line but I also want the colors to be the same for each item in the group.    I thought that I could exclude the color from the lineattrs so that the defaults would just be used for each of the series but it's not working the way I had hoped. I've tried a lot of the suggestions online but still not working. Originally I want to do this with two by variables (drug and prodname) but that was making it more complicated. I ended up wrapping the plot in a macro and calling out each drug and product name to plot separately.   The following is an hypothetical example with a drug called 'abcdrug' and prodname is blank. Is there a simple statement I can add or ods graphics option I can use without creating numerous lines of code? I often want to do these sorts of graphs on the fly and sometimes not worth creating all the code for it.    %macro shareplot(dname=, prod=) ; proc sgplot data=share (where = (drug="&dname." and prodname="&prod.")) ; title "Shares Over Time"; title2 "&dname., &prod." ; series y=share x= month/ group=strength lineattrs=(thickness=2 pattern=solid) legendlabel="Share by Month" ; series y=share_12 x= month/ group=strength lineattrs=(thickness=2 pattern=longdash) legendlabel="Share by Month - 12 month MA" ; YAXIS LABEL = 'Share (%)' ; YAXIS min=0 max=1 ; FOOTNOTE " "; where MONTH>='01JAN2013'd ; run; %mend plot ; %shareplot(dname=abcdrug); ... View more

I am trying to create a small database of months from one date to another, incrementing by month. I want three observations: 10/1/2017 11/1/2017 12/1/2017   The following code just gives me 10/1/2017 and doesn't loop.   Assume: %Let maxmth=01SEP2017; %Let maxdt=01DEC2017;   data addMonths ; format month MMDDYY10. ; month=intnx('month',%unquote(%str(%')&maxmth.%str(%'d)),1) ; put month ; DO i = (month <= %unquote(%str(%')&maxdt.%str(%'d))) ; output ; month=intnx('month',month,1) ; put month; END; run; proc print ; run; ... View more

I ran this code all day yesterday without an issue. Today it won't run and I am getting this error message. I've seen messages on SAS support related to this issue but they were from 2014 and a paper written in 2012 suggested adding an extra semi-colon which did not work. I am using SAS Base 9.4 and the path is a network path and is correctly written and the file exists (again, worked all day yesterday). Does anyone know why this could be happening with SAS 9.4. None of my %put statements are resolving either. It seems like SAS is just malfunctioning its macro utility.   4663 %include "&path.Plot Workbook.sas" ; ERROR: Incorrect %INCLUDE statement will not be executed. There is a syntax error. 4664 %include "&path.Plot Workbook.sas" ;; ERROR: Incorrect %INCLUDE statement will not be executed. There is a syntax error. ... View more