Choice  the best estimations of parameters robust regression model with practical application
A Thesis Submitted to
Council of The college of Administration and Economics / Karbala university 
As  Partial  fulfillment   of the Requirements for the Degree of master s
in Statistics of Science
Researcher
 Saif Hussein Ali
Supervised By
Dr. Assistant prof. Shrooq Abd-AL Rida Saeed

ABSTRACT

     Mostly the statistical methods depend on hypothesis that probabilitical distribution for data of regression model is normal , but in applied state at most to be this data have other distributions because of the extreme values , in order to use methods are not sensitive to extreme values . and give efficient estimations such as robust estimation methods .  

  In this thesis study seven methods have been estimated the parameters of robust multiple regression model ، Two methods of classical methods ، and five methods are robust  methods ، Because of the presence of effect of extreme values , It has been drawn simple random sample (151) patients infected with infertility represented by dependent variables (egg size) and independent variables (age , weight , infertility kind , L.H hormone , F.S.H. hormone , P.L. hormone ) .

to aim compare between these methods in order to find the best method for estimation by using measures reflect the quality and efficiency of those estimates , such as mean square error (MSE) and determination facter ( ) to messure the model efficiency ، by using statistical program (stata) ، The results show that the method of M, Tukey’sbiweight is the best and secondly method is M, Andrews sine، While S is the worst among of all robust method ، Either the classical methods has demonstrated its failure to estimat efficient  estimators . Because of presence of extreme values.