An Estimation Of The Parameter Based On Floating-point Binary Data Rounding

In statistical analysis,many problems of statistics are solved with the help of computers.Data is usually assumed to be real number. But due to the limited memory required of thecomputer, Data is often through a certain degree of approximation. This paper mainly dis-cussed the analysis after rounded the data. Because the numbers on the computer are writtendown in binary codes and round-of in floating-point, this paper funded a way to solve thisproblem. In particular, we discussed how to estimate the parameters if the data were got fromthe rounding samples of the independent and identically distributed distribution or linear modelwith unknown parameters. The estimators of the parameters that obtained by using the max-imum likelihood estimation method are much closer to the true values, holds the consistency,asymptotic normality and other good properties as well.