High Breakdown Weighted LAD Regression Estimator
| المصدر: | مجلة القادسية للعلوم الإدارية والاقتصادية |
|---|---|
| الناشر: | جامعة القادسية - كلية الادارة والاقتصاد |
| المؤلف الرئيسي: | Uraibi, Hassan S. (Author) |
| مؤلفين آخرين: | Alwan, Hatem Abd A. (Co-Author) |
| المجلد/العدد: | مج24, ع1 |
| محكمة: | نعم |
| الدولة: |
العراق |
| التاريخ الميلادي: |
2022
|
| الصفحات: | 405 - 409 |
| ISSN: |
1816-9171 |
| رقم MD: | 1269663 |
| نوع المحتوى: | بحوث ومقالات |
| اللغة: | الإنجليزية |
| قواعد المعلومات: | EcoLink |
| مواضيع: | |
| كلمات المؤلف المفتاحية: |
Weighted LAD | Masking and Swamping | Robust Location and Scale Estimators
|
| رابط المحتوى: |
الناشر لهذه المادة لم يسمح بإتاحتها. |
| المستخلص: |
The least Absolute Deviation estimator has a high breakdown point probably reach 50%, so it provides a good alternative to the LS estimator when vertical outliers are present in the data set. However, It may lose this feature when the design matrix x is having at least one leverage point. Many authors have been efforts in the literature to assign a down weight to leverage point and suggested the weighted LAD regression, which is denoted as WLAD for increasing the breakdown point of LAD. Most weighted functions that are discussed in the literature are based on robust Mahalanobis distance, which is a familiar approach to identifying leverage points. Unfortunately, this robust distance could be affected in appearing high leverage points or when the masking and swamping phenomena have happened. Consequently, robust Mahalanobis distance may not detect all leverage points, and then WLAD would be a non-robust method, and its estimator surely will break down. In this paper, we proposed improving Mahalanobis’s distance based on weighted fast and consistent high breakdown estimator location and scale matrix, which may make WLAD is more robust than the previous ones. The simulation studies have been done and the results of our proposed UWLAD is compared with AWLAD and GWLAD methods which well known in the literature. The result shows that the performance of the UWLAD method is more efficient and reliable than others. |
|---|---|
| ISSN: |
1816-9171 |
عناصر مشابهة
-
Bayesian Estimation for Semiparametric Logistic Regression
بواسطة: Alshaybawee, Taha Hussein منشور: (2022) -
The Multilinearity with Principal Component Regression
بواسطة: Al-Husseini, Fadel Hamid Hadi منشور: (2022) -
Linear Possibility Model for Ordered Categorical Data: A Similar Way of Analysis to Regression Analysis
بواسطة: Adam, Amin I. منشور: (2015) -
Estimating the Linear Regression Model in High-Dimensional Data and Collinearity
بواسطة: Al Qadhi, Sahar Abd Almuain Hassan منشور: (2020) -
Bayesian Estimation for Single Index Quantile Regression with Normal Scale Mixture
بواسطة: Alshaybawee, Taha Hussein منشور: (2019)
