An approximate numeric type based on IEEE 754. It isn't a good fit for money, rounding or testing equality; decimal is the exact type to use in those cases.
Also called real is float(24).
Read more: Microsoft Learn
In the Ultra Transcenders books
Each book explains float in context, with comparison tables and the common traps.
Terms in this definition
- IEEE 754
The floating-point standard used by Power BI's Decimal number type. Because values are stored approximately, comparing for equality, ranking with RANKX or adding up large mixed numbers can behave oddly; Whole number or Fixed decimal number sidestep the problem.
- DECIMAL
Use
DECIMAL(p,s)for money or any figure where floating-point rounding would be wrong. Precisionpcan be 1 to 38 and is 10 if left out; scalesis 0 if left out.
Related terms
- float
A
FLOATtakes up 4 bytes and has single precision. When exactness counts, convert it toDECIMAL, as totals and averages of it may otherwise make a materialized view refresh in full. - Measure table
Created via Enter data, this table holds nothing but measures; hiding its one column makes it float to the top of the Data pane.