In this merged data, there is a column with missing values. For those values, I want to compute the mean of that column, and then impute that mean for the missing values. I'm trying to compute the mean and make it a global macro variable:   There is no need for a macro variable here. SAS has many PROCs to do what are standard statistical methods, which you should investigate (or ask us about) before you decide to create your own code to do a standard statistical method, and before you decide to create macro variables. As @Ksharp points out, PROC STDIZE will do this for you. ... View more

the NORMAL function is deprecated (that means it is an old function and best practice is that you should not use it any more) https://documentation.sas.com/doc/en/pgmmvacdc/9.4/lefunctionsref/p05tjojnyjll0zn1qq3xjl9aawl0.htm  The 999 was a "seed" that you could use to make the sequence of random numbers generated reproducible.   Better to use the rand('normal') function and CALL STREAMINIT() to specify a seed and make the random numbers reproducible.   ... View more

You don't even need to do this in SAS, you can do this in your head.   Add 60 to the birth year   2013+60=2073   But maybe that gives you an idea about how to do it in SAS ... View more

@raja777pharma wrote:   Can any one suggest how to populate only year and month values. You want to take the first 7 characters of dtc_ and ignore the rest. You will also need to remove the hyphen from these dates.   dtc1_=compress(substr(dtc_,1,7),'-'); dt=input(dtc1_,yymmn6.); format dt yymm6.; /* or any other format you want */   There is no informat YYMMD which is why you get the error. There is an informat YYMMDD (with two Ds on the end), but that won't work on the data in DTC_ that you show.   You might want to talk to whomever provides you with dates as character strings that this is not a good thing to do. Or if you converted the dates to character strings earlier in the program, don't do that. Keep dates as numeric variables.   Please do not provide the log as a screen capture. Please provide the log by copying it as text and then pasting it into the window that appears when you click on the </> icon. ... View more

Your 4.1 in an INPUT statement is an INformat, not a format. It tells SAS how to read the numbers. If you want to tell SAS how to display (or write to an output) the numbers, then you need a format. This allows numbers to read in a certain way, and displayed (or written to an output) differently, if so desired.   data a; input source 4.1; format source 4.1; cards; ... ;   ... View more

Agreeing with @ballardw    Don't write to a .LST file when you can write to a SAS data set. ... View more

I am not setting properly the options.     Explain this. Why do you think this? If you are getting errors, please show us the LOG. If you are getting the wrong results, show us the incorrect results and explain what you think the correct results are. Please don't make us guess, please don't make us re-do the work you have already done. ... View more

@altadata1 wrote: Thank you for your response @PaigeMiller. In my case, I'd like to create two new variables like this:  when tag is between A00 and B90, if tag equals to A23, B19, B21, then new_var=1, the rest new_var_2=2.  Many thanks again.  This is a relatively simple change to the code I provided, and I leave it to you as a homework assignment to make the change.   Since these two conditions are mutually exclusive, I recommend putting both conditions into one single variable, as I did, rather than two variables, unless you have a very good reason why it needs to be in separate variables. If there are two and only two values that a variable can take on, 0 and 1 is recommended and preferred over 1 and 2. ... View more

You have to code the DATA step such that tag is a character variable, and so the values of tag must be in quotes. If you don't you will get errors. So, A00 should be 'A00' in your code, and so on.   data want; set a; if 'AOO'<=tag<='B90' then do; if tag in ('A23','B19','B21') then new_var=1; else new_var=0; end; run; ... View more

I don't think @Ksharp has provided the best answer. First, PROC SUMMARY will be a lot less typing and probably execute faster. In the case where your real data set has more than the three variables (age sex race), you will have to do a lot more typing in SQL than you would have to do in PROC SUMMARY.   If there is a need to slice the data by two (or more) variables at one time, such as all combinations of age and sex, PROC SUMMARY handles this easily, with little extra coding. The extra coding needed for an SQL solution is not obvious.   Furthermore, if the real data has missing values in variable COUNT, SQL gives the wrong answers, while SUMMARY gives the right answers.   PROC SUMMARY is a much better tool here. ... View more

This looks very close to me, small changes needed to PROC SUMMARY. You need a data step after PROC SUMMARY to do the actual division from the results in data set TOTALS.   proc summary data=____________ nway; class county; var opioidepisodes totalepisodes; output out=totals sum=; run; data totals1; set totals; rate=opioidepisodes/totalepisodes; run;     PS: When you want to show us your code, copy it as text and paste it into the window that appears when you click on the "little running man" icon. Making screen captures of text is not a good thing to do. ... View more

Honestly, a larger data set (covering a whole year) even if it is fake data, and showing us the desired output would help us understand the problem better.   It sounds like you want some sort of sum or average or weighted sum or weighted average over the entire year. PROC SUMMARY will do this. It can also easily slice the data into different results for combinations of age sex and race.   ... View more

If you want a SAS data set with different formats on each row, I agree with @Kurt_Bremser it can't be done.   However, if you want a REPORT, then it is possible in PROC REPORT. But of course this violates the "do everything in PROC SQL" rule and also that you want the results as a SAS dataset.   So if you're desperate, I'm afraid you are going to remain desperate. Perhaps you can convince your client that sometimes SQL can't do everything.   You could use character variables, but of course then you can't perform arithmetic (like addition) or finding maximums from these character variables. (And then, it sounds like you want a report, not a data set) ... View more

Please show us a portion of your input data set. Please use these examples and instructions to provide the data as working SAS data step code. Please do NOT show us screen captures of data or provide data as Excel files.   Please also show us the desired output. This can be a screen capture if you wish. ... View more

In general, I am skeptical that this is a good approach. Can you please share with us what you will do with these macro variables once they are created, that cannot be done using data set HAVE?   Anyway, here is the code you want.   data _null_; set have; call symputx(cats('num_val',_n_),num_val); run; ... View more