@runn wrote: I think this is close to solution but It's not giving me the desired outcome. The newly created want have zero observations even on imputing correct have No idea what that means. You want the percent of 1, in mathematical terms that's the SUM(N=1)/ Sum (N=1) + SUM(N=0) => using some math that works out to the same average value of 0/1. Make sure to remove the HAVING clause which filters out the results to include only percentages greater than 50, which was your original question.    You could also use a plain proc means and the average would be the percentage as well.  ... View more

@runn wrote: I appreciate Reeza but I have 1.5 million observation. It makes SAS slower if I past all the observations as a dateline in the code.   There should be some other way as well.   Thanks. Why would you post the data lines in the code???? If you have the data in a text file then use a INFILE statement to tell SAS where to read the data. Reeza's example is just to show you how your INPUT statement should look based on your description of your record layout. ... View more