The College of Administration and Economics at the University of Baghdad discussed , a master’s thesis in field of Statistics by the student (Farah Emmad Mohammed) and tagged with (Estimation of Multiple Log-Gamma Normal Regression Model in Case ASymmetric Data Application) , Under supervision of (Prof Dr. Qutaiba Nabeel Nayef )
In this research, the multiple linear regression model with errors following the Log-Gamma-Normal distribution was studied, which is one of the methods for modeling asymmetric data and estimating its parameters using three estimation methods: Maximum a Posteriori (MAP), Generalized Maximum Entropy (GME), and Maximum Likelihood (ML). In addition, some basic concepts related to the thesis were presented, including outliers, symmetric data, asymmetric data, skewness and kurtosis coefficients, and Log-Gamma-Normal distribution characteristics. The simulation was conducted with sample sizes (25, 50, 75, 100) and default values for the LGN regression model parameters (β, σ, τ, δ). The results showed the superiority of the MAP method over the rest of the estimation methods because it has the lowest RMSE, AIC, and BIC, and it has the property of stability and non-fluctuation with changing sample sizes. It is confirmed that it is an improvement in estimation of the maximum likelihood ML method because it employs the initial distribution around the parameters as one of the Bayesian estimation methods.
