proc means for subcategories of data

[6071]

Hello,

I have 90 lines of data.  I want to calculate the mean and some other metrics for the first 35 years of data, then for years 2-36, then 3-37, etc and then view all of these mean series.  How do I do this?  Thanks very much for your help.

[PeterClemmensen]

Can you show us your data?

[PaigeMiller]

One method is using a multilabel format

Example:

proc format;
    value yearf (multilabel) 
        1-35='1-35'
        2-36='2-36'
        3-37='3-37'
        /* I'm lazy, you type the rest */
;
run;

proc summary data=have nway;
    class year_since_start/mlf format=yearf.;
    var some_other_variables;
    output out=_stats_ mean= std= min= max=/autoname;
run;

Another method, if you have SAS/ETS and PROC EXPAND, is to calculate moving averages of length 35 based upon your year variable. Probably less typing.

[Reeza]

Moving average or rolling average is the term to use in Google for this, but it depends on how your data is structured.
If you have SAS/ETS this is quite easy, but also very doable within a data step as well.

SAS/ETS example of 12 month moving average
https://gist.github.com/statgeek/07a3708dee1225ceb9d4aa75daab2c52

Data step, array approach
https://gist.github.com/statgeek/27e23c015eae7953eff2