Why can’t I compare two values that I know are equal?
| Title | Comparing floating-point values (the float function) | |
| Author | Jeremy B. Wernow, StataCorp |
If you try to compare the value of a floating-point variable to a known value, you may not get the results you expect.
. list
+-----+
| x |
|-----|
1. | 7.3 |
2. | 7.3 |
3. | 7.3 |
+-----+
. list if x==7.3
. describe
Contains data
obs: 3
vars: 1
size: 24 (99.9% of memory free)
-----------------------------------------------------------------------
storage display value
variable name type format label variable label
------------------------------------------------------------------------
x float %9.0g
------------------------------------------------------------------------
Sorted by:
Note: dataset has changed since last saved
This occurs because of a binary rounding error. There is no way to exactly represent most noninteger numbers using a binary system. To get around this, you can use the float() function (not the cast type). To round the above expression to its floating-point value, type
. list if x ==float(7.3)
+-----+
| x |
|-----|
1. | 7.3 |
2. | 7.3 |
3. | 7.3 |
+-----+
To read more about floating-point precision, see [U] 13.12 Precision and problems therein.
