Prof. Bassem Shaleba Muslim, Prof. Dr. Seif El-Din Hashem Qamar

JOURNAL OF ADMINISTRATION AND ECONOMICS

Study the Properties of Generalized Maximum Entropy through of Estimate the Four-Parameter Weibull Growth Model

In this research, the characteristics of the Generalized Maximum Entropy method were studied when estimating the parameters of the four-parameter Weibull growth model using simulation according to the requirements of the method (the dimensions of the support variables vector, and the values of the constant C) as well as the type of random error distribution and the sample size. Depending on the comparison scale represented by the mean error squares MSE for the studied phenomenon model, optimal cases were simulated. The data were generated according to the standard normal distribution and t distribution with (3) degree of freedom, and Chi-square distribution with (4) degree of freedom, with the sample size (10, 20, 30), as well as assuming the constant C of (1, 3, 5, 7, 9) and the support variables vector M of (5, 7). The simulation results showed the effect of the normal distribution and the Chi-square distribution with degree of freedom of (4) by changing the dimension of the support variables vector between (5, 7) on choosing the value of the constant C for all sample sizes. As at M = 5 it is preferable to choose it from the period (3, 5) for the sample size (10, 20) and begins to decrease as the sample size increases, but at M = 7, the choice is opposite to the case of M = 5 in relation to the standard normal distribution, whereas for the Chi-square distribution with a degree of freedom (4) it is preferable to choose a large value for the sample size (10) and a value from The period (3, 5) for the sample size (20, 30), but in the case of t distribution with a degree of freedom (3), there is no effect for choosing the value of the constant C by changing the dimension of support variables vector, and it is preferable to choose the value from the period (5, 9) and this value increases with increasing sample size.