DATA Step, Macro, Functions and more

Dealing with E values

Reply
N/A
Posts: 0

Dealing with E values

Hi-
I'm doing a proc compare on two different dataset and wanted to get a resulting dataset with just the records with different values. But for some reason even though I have the same values the diff is calculated as a E value ( eg: 28.4-28.4=3.552714E-150). I know this is a place for data step questions but this happens even if I do a difference in the datastep. As of now to keep rolling I created an array that looks at each record and variable with "E" and changed to "0". But is there a way/function that I need to incorporate to over come this issue.

Regards,
Addy
Regular Contributor
Posts: 165

Re: Dealing with E values

Floating point error is annoying but it happens. Use C=fuzz(A-B) .
Super User
Posts: 3,102

Re: Dealing with E values

What COMPARE method are you using? If you use EXACT it is likely to report the floating point errors you are finding. Try ABSOLUTE is it should ignore small differences.
Ask a Question
Discussion stats
  • 2 replies
  • 112 views
  • 0 likes
  • 3 in conversation