The development in manufacturing computers from both (Hardware and Software) sides, make complicated robust estimators became computable and gave us new way of dealing with the data, when classical discriminant methods failed in achieving its optimal properties especially when data contains a percentage of outliers. Thus, the inability to have the minimum probability of misclassification. The research aim to compare robust estimators which are resistant to outlier influence like robust H estimator, robust S estimator and robust MCD estimator, also robustify misclassification probability with showing outlier influence on the percentage of misclassification when using classical methods. ,the other aim of research is to compare estimators to find the best estimator which can gave less probability of misclassification especially with the variety of contamination percentage and different samples sizes and the data contaminated according to a technique that had never been used in other research on the country level.
1.jpg)