مقارنة بعض الطرائق الحصينة في تحليل الارتباط القويم بإستخدام المحاكاة مع تطبيق عملي

Abstract

The important of the necessity to my efficient  estimation methods, which are called the robust methods, appears when the data of the studied phenomenon are contaminated, it means the observations contains outliers,which may produce estimators which result in   increasing (decreasing) in the (MSE), That would leads to an inaccurate statistical inference.

From this point was the good behind this research in reaching robust estimators of canonical correlation analysis that can be achieved through the study of some robust methods such as (estimators – M, estimators – MVE, estimators-MCD and estimators-S) . which tow types of applied studies were made, The first one uses the simulation method to compare among the studied methods of estimation in canonical correlation analysis and detect the best of the estimator depending on the two statistical measurements bias mean and mean square error which renders the minimum MSE of canonical correlation analysis, the second study uses the truthful data to verity the performance in a practical actuality.

In general it is regarded that MCD estimator is the best in the estimation of canonical correlation analysis in comparison with the other studied methods of estimation.

The important of the necessity to my efficient  estimation methods, which are called the robust methods, appears when the data of the studied phenomenon are contaminated, it means the observations contains outliers,which may produce estimators which result in   increasing (decreasing) in the (MSE), That would leads to an inaccurate statistical inference.

From this point was the good behind this research in reaching robust estimators of canonical correlation analysis that can be achieved through the study of some robust methods such as (estimators – M, estimators – MVE, estimators-MCD and estimators-S) . which tow types of applied studies were made, The first one uses the simulation method to compare among the studied methods of estimation in canonical correlation analysis and detect the best of the estimator depending on the two statistical measurements bias mean and mean square error which renders the minimum MSE of canonical correlation analysis, the second study uses the truthful data to verity the performance in a practical actuality.

In general it is regarded that MCD estimator is the best in the estimation of canonical correlation analysis in comparison with the other studied methods of estimation.