condition if var=number with two decimal place

[Angela Jiang]

Hi,

I have a variable displayed like below, It is of type float and format %9.0g. There are two numbers that occurs with more frequency than others, 620.32 and 685.62. I would like to generate a new variable that equals to 1 if var1 takes 620.32 or 685.62. However these numbers actually have more precision than what is displayed. I am struggling with how to do this in a easy way (my current code is gen category=1 if inrange(var1,620.31, 620.33) | inrange(var1,685.61, 685.63)). I tried rounding and recasting the variable as double but the precision doesn't seem to decrease. It would be great if you can suggest a quick way to do this. Thanks!

var1
rate_06
620.32
620.32
620.32
620.32
620.36
620.54
620.54
620.57
685.62
685.62
685.62
685.62


Best,
Angela

[Rich Goldstein]

h precision

which both explains the issue and offers solutions

[Mike Lacy]

An old maxim of computer management of data was and is "Never compare floating point numbers for exact quality." (This was on the wall of a university computer room when I was a student, and I had to struggle with a situation like the current one before I understood it.) I suppose there are situations in which this maxim could or should safely be violated.

[Angela Jiang]

Thanks Rich and Mike!