As per the float and real (Transact-SQL) float is an approximate data type.
But also in the same article it is being mentioned that float has precision of 15 digits.
What exactly it mean by Precision of 15 digits?
It means that the first 15 digits are approximately correct. Thanks to IEEE 754 there are many numbers that can't be exactly represented, but you can be sure that the first 15 digits are very close.