About LaurieF

LaurieF · ‎02-03-2017

Without seeing the rest of your data, it's hard to give a hard and fast answer. This will work for the example you've given, however: data _null_; length address $ 50; address = 'POST 1234 ST NO 50.'; address = tranwrd(address, ' ST ', ' STREET '); put address=; run; tranwrd changes character strings, whereas translate changes single characters. Note that I have prefixed and suffixed the two strings with spaces, because POST would have been changed as well. However you may strike problems with ST., where a full-stop is after the string - you may have to put in two tranwrd calls to cope with this. I have a very large and complex macro which does exactly this, but it may be more than you need. Note that the parameter order for tranwrd is different to translate (which only changes single characters): the second parameter is what you want to change, the third is what you want it to change to.

LaurieF · ‎02-03-2017

Well, here's another way of doing it, which doesn't require scan at all, relying on the wonderful internal machinery of the infile statement: data products; infile cards dsd dlm=' '; length product $ 30; input product @@; if product ne 'and' and not missing(product); keep product; cards; orange apple juice and cola coffee candy and paper banana sugar and pen and book ; run; You could stuff around for ages playing with buffer pointers, but if your requirements are as simple as you present, this may be all you need. It also gets around the problematic cANDy!

LaurieF · ‎02-03-2017

Simplify, simplify, simplify: %macro date_loop(start,end); %do i = 0 %to %sysfunc(intck(month, "&start"d, "&end"d)); %let date = %sysfunc(intnx(month,"&start"d, &i, b), date9.); %put &date; %end; %mend date_loop; %date_loop(01jul2015, 01feb2016);

LaurieF · ‎02-02-2017

Just not today! But tomorrow I probably will.

LaurieF · ‎02-02-2017

No, multiple %sysfuncs are fine. It can make it hard to read, but it does work very well.

LaurieF · ‎02-02-2017

But - but - in your intnx function in the %let within the %do loop - you're passing in 01jul2015, not 20270. You have to convert it somewhere! I have never, ever, made this mistake mistake. Oh, no, no, no, no, no… BTW, you can convert dates like that on the fly with %sysevalf, funnily enough. %let date=%sysfunc(intnx(month, %sysevalf("&start"d) ,&i,b),date9.); or even simply %let date=%sysfunc(intnx(month, "&start"d ,&i,b),date9.);

LaurieF · ‎02-02-2017

I'm with @RW9 - upgrade, if you're in the position to do it. If you can't, stick with multiple index or indexw calls, which will work fine.

LaurieF · ‎02-02-2017

I can't tell. It should work. But as I said, let's see your data and we'll find out.

LaurieF · ‎02-02-2017

I don't have access to your source data, but when I run data work.systems; set sashelp.cars; if prxmatch('m/Acura|Quattro|Ford/oi',model) > 0 then found=1; else found=0; run; it works fine. If you want to get fast accurrate solutions, posting the data you're using is recommended. It's fortunate there's a similar dataset in sashelp.

LaurieF · ‎02-01-2017

Probably not. Can you show your code?

LaurieF · ‎01-31-2017

I was trying to say d__n, by the way, in case you thought I was aiming for something stronger.

LaurieF · ‎01-31-2017

Oh **bleep** - I misremembered. I apologise.

LaurieF · ‎01-31-2017

Nah - data step the same. I like to be explicit, because it makes the code less ambiguous. But it's not necessary. It doesn't work the other way though, of course!

LaurieF · ‎01-31-2017

Oh it works fine, but I agree, it should be coalescec. 62 proc sql; 63 create table class_coalesce as 64 select coalesce(name, 'xxx') as name length=30 65 from sashelp.class; NOTE: Table WORK.CLASS_COALESCE created, with 19 rows and 1 columns.

LaurieF · ‎01-31-2017

It's not very well formed, but I'll give it ago. (Translated: I wouldn't have written it quite like that) Your master table, inp_1, contains at least three columns: mbr_id, qualifier and dx. It is being joined with testtable (I hate tables called that - sorry if it's yours, but I really do!) which contains icd10, icd9 and smm. The target table will contain inp_1's mbr_id and either icd_9 or space (where icd_9 is missing or the join failed) if qualifier is 2, or dx. The where clause is possibly problematic. If a join against testtable can't be found, it will always be true because of the coalesce: smm will be missing, so the function will return 0. However if the join is successful and smm is still missing, again it will be true. The group by is wrong too: there's so grouping of variables. Turn that into an order by. The SQL processor will return a warning message to this effect. But without seeing the data, I can only ever guess.

Online Status	Offline
Date Last Visited	‎09-02-2025 04:43 PM