Use the approximate numeric types, float, double precision, and real, for numeric data that can tolerate rounding during arithmetic operations. The approximate numeric types are especially suited to data that covers a wide range of values. They support all aggregate functions and all arithmetic operations except modulo.
