MODIFIED FUZZY-ROBUST RIDGE REGRESSION FOR MULTICOLLINEAR, OUTLIER-CONTAMINATED DATA

Vaman M Salih; Shelan S Ismaeel

Article Details

MODIFIED FUZZY-ROBUST RIDGE REGRESSION FOR MULTICOLLINEAR, OUTLIER-CONTAMINATED DATA

Authors:

Vaman M Salih | Shelan S Ismaeel

Publisher:

University of Zakho, Iraq

Journal:

Science Journal of University of Zakho Vol.14 Issue 2, pp.209-219

Document Type:

Article

Publication Date:

4/2026

Document Language:

English

Document ID (Org No.):

2001935088

pISSN:

2663-6298

Keywords:

Abstract:

 Multicollinearity is known to have a significant impact on the stability of linear regression parameter estimation, while the presence of outliers tends to compound this problem. Ridge regression helps to improve the multicollinearity problem, but it is highly sensitive to outliers. This paper proposes Modified Fuzzy Robust Ridge Regression (MFRRR), which modifies classical ridge regression by adapting the penalty parameter through modified fuzzy robust estimators based on weighted residual membership functions. The method is evaluated under challenging data conditions involving simultaneous multicollinearity, outliers, and fuzzy uncertainty. Performance is assessed using both a real body fat dataset and Monte Carlo simulations with varying sample sizes  correlation levels   and contamination rates  . MFRRR is compared to ordinary least squares (OLS), ridge regression, and robust ridge regression based on the mean absolute error (MAE) as an evaluation criterion. These findings indicate that MFRRR is always associated with smaller prediction errors and more reliable parameter estimates, especially when there is high multicollinearity and data contamination.

Article Details

MODIFIED FUZZY-ROBUST RIDGE REGRESSION FOR MULTICOLLINEAR, OUTLIER-CONTAMINATED DATA

Authors:

Vaman M Salih | Shelan S Ismaeel

Publisher:

University of Zakho, Iraq

Journal:

Science Journal of University of Zakho Vol.14 Issue 2, pp.209-219

Document Type:

Article

Publication Date:

4/2026

Document Language:

English

Document ID (Org No.):

2001935088

pISSN:

2663-6298

Statistics and Data Analysis

Fuzzy sets

Multicollinearity

Outliers

Ridge Regression

Robust regression

Fuzzy robust regression

Issues

Article Details

Authors: Vaman M Salih | Shelan S Ismaeel

Publisher: University of Zakho, Iraq

Journal: Science Journal of University of Zakho Vol.14 Issue 2, pp.209-219

Document Type: Article

Publication Date: 4/2026

Document Language: English

Document ID (Org No.): 2001935088

pISSN: 2663-6298

Issues

Authors:

Vaman M Salih | Shelan S Ismaeel

Publisher:

University of Zakho, Iraq

Journal:

Science Journal of University of Zakho Vol.14 Issue 2, pp.209-219

Document Type:

Article

Publication Date:

4/2026

Document Language:

English

Document ID (Org No.):

2001935088

pISSN:

2663-6298