Comparison of the percentile estimation method and mixture (maximum likelihood and least square) method for estimating parameters of Johnson bounded distribution

Naeem, Sanaa Mohammed; Hussein, A. N.

doi:10.22075/ijnaa.2022.5987

Comparison of the percentile estimation method and mixture (maximum likelihood and least square) method for estimating parameters of Johnson bounded distribution

Document Type : Research Paper

Authors

Sanaa Mohammed Naeem

A. N. Hussein

College of Management and Economics, University of Basrah, Iraq

10.22075/ijnaa.2022.5987

Abstract

The Johnson Bonded Distribution (JSBD) is one of the distributions belonging to the Johnson Distribution family (JD) This family is considered one of the flexible distributions that enable it to represent the random behaviour of many phenomena. The method of percentile estimation and the Mixture method (Maximum Likelihood & Least Square) were used to find estimates of the distribution parameters. The simulation results showed the advantage of the Mixture method at large sample sizes, the accuracy of the two methods is close at average sample sizes.

Keywords

Johnson distribution

Johnson bounded distribution

percentile method

Mixture method

[1] M.R. Flynn, Fitting human exposure data with the Johnson SB distribution, J. Exposure Sci. Environ. Epidemio. (16) 2006 56–62.

[2] F. George, Johnson’s System of Distributions and Microarray Data Analysis, Theses, University of South Florida, 2007.

[3] D.T. Mage, An explicit solution for SB, parameters using four percentile points, Technomet. 22(2) (1980) 247–251.

[4] P. Palimitesta and C. Provasi, Estimating distribution functions in Johnson translation system by the starship procedure with simulated annealing, Int. J. Stat. 0 (1-2) (2000) 97–110.

[5] A.T. Soyink, On the parameter estimation of Johnson’s system of distribution, Benin J. Stat. 3(28) (2020) 82–90.