تعديل اختبار كولد فيلد-كوانت المعدل لتشخيص مشكلة عدم التجانس بوجود الشواذ

Abstract EN

Rana et.

al.

in 2008 proposed modification Goldfield-Quant test for detection of heteroscedasiticity in the presence of outliers by using Least Trimmed Squares (LTS).

The proposed method assumed trimming ten percent from dataset.

In this situation may take place losing in clean data where the real percentage of outliers is less than (10%) or the estimator may be misleading where the outliers percentage is more than (10%).

The researcher propose another modification of Goldfield-Quant test to be robust.

The results show our proposed method is better than the last modification of Goldfield-Quant test