The aim of this research is to estimate the parameters of the linear regression model with errors following ARFIMA model by using wavelet method depending on maximum likelihood and approaching general least square as well as ordinary least square. We use the estimators in practical application on real data, which were the monthly data of Inflation and Dollar exchange rate obtained from the (CSO) Central Statistical organization for the period from 1/2005 to 12/2015. The results proved that (WML) was the most reliable and efficient from the other estimators, also the results provide that the changing of fractional difference parameter (d) doesn’t effect on the results.

   Nonlinear regression models are important tools for solving optimization problems. As traditional techniques would fail to reach satisfactory solutions for the parameter estimation problem.  Hence, in this paper, the BAT algorithm  to estimate the parameters of  Nonlinear Regression models is used . The simulation study is considered to investigate the performance of the proposed algorithm with the maximum likelihood (MLE) and Least square (LS) methods. The results show that the Bat algorithm provides accurate estimation and it is satisfactory for the parameter estimation of the nonlinear regression models than MLE and LS methods depend on Mean Square error.

Abstract

Characterized by the Ordinary Least Squares (OLS) on Maximum Likelihood for the greatest possible way that the exact moments are known , which means that it can be found, while the other method they are unknown, but approximations to their biases correct to 0(n^-1) can be obtained by standard methods. In our research expressions for approximations to the biases of the ML estimators (the regression coefficients and scale parameter) for linear (type 1) Extreme Value Regression Model for Largest Values are presented by using the advanced approach depends on finding the first derivative, second and third.

In this paper we estimate the coefficients and scale parameter in linear regression model depending on the residuals are of type 1 of extreme  value distribution for the largest values . This can be regard as an improvement for the studies with the smallest values . We study two estimation methods ( OLS  & MLE ) where we resort to Newton – Raphson (NR) and Fisher Scoring methods to get MLE estimate because the difficulty of using the usual approach with MLE . The relative efficiency criterion is considered beside to the statistical inference procedures for the extreme value regression model of type 1 for largest values . Confidence interval , hypothesis testing for both scale parameter and regression coefficients

In this paper we used frequentist and Bayesian approaches for the linear regression model to predict future observations for unemployment rates in Iraq. Parameters are estimated using the ordinary least squares method and for the Bayesian approach using the Markov Chain Monte Carlo (MCMC) method. Calculations are done using the R program. The analysis showed that the linear regression model using the Bayesian approach is better and can be used as an alternative to the frequentist approach. Two criteria, the root mean square error (RMSE) and the median absolute deviation (MAD) were used to compare the performance of the estimates. The results obtained showed that the unemployment rates will continue to increase in the next two decade

The purpose behind building the linear regression model is to describe the real linear relation between any explanatory variable in the model and the dependent one, on the basis of the fact that the dependent variable is a linear function of the explanatory variables and one can use it for prediction and control. This purpose does not cometrue without getting significant, stable and reasonable estimatros for the parameters of the model, specifically regression-coefficients. The researcher found that "RUF" the criterian that he had suggested accurate and sufficient to accomplish that purpose when multicollinearity exists provided that the adequate model that satisfies the standard assumpitions of the error-term can be assigned. It