The Redescending M estimator For detection and deletion of Outliers in Regression analysis

Anekwe, Stella; Onyeagu, Sidney I.

doi:10.18187/pjsor.v17i4.3546

Cited by 4 publications

(4 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The sub-panel namely Assigned value computes the assigned value and the standard deviation for proficiency assessment through consensus value among participating results. This sub-panel contains the following estimation methods: Horwitz curve, algorithm A (Huber's M-estimator), algorithm S [2], Tukey's biweight M-estimator [13] and Hampel's three-part M-estimator [14]. The second sub-panel (Scores) contains the performance statistics z score, z 'score, zeta score, E n score [2], Q-scoring, QM E R (Relative Quadratic Mean Error), corrected z score and robust z score [4].…”

Section: Univariate Modulementioning

confidence: 99%

WAPT: a friendly interactive web application for performance assessment in interlaboratory comparisons

Leal,

Rocha

2024

Preprint

View full text Add to dashboard Cite

A free and user-friendly web application for performance assessment in interlaboratory comparisons (IC) was developed to aid IC providers with statistical analysis of results reported by participants. WAPT has three modules that can deal with univariate, categorical, and multivariate data sets. Each module provides well-established statistical methods and recent approaches to proficiency testing assessment. This brief communication presents some features of WAPT using data sets examples from: (i) proficiency testing to evaluate the performance of food and beverage analysis laboratories in the determination of benzoic acid in orange juice; (ii) proficiency testing to verify the ability of participating laboratories to correctly identify 19 salmonella serotypes; (iii) interlaboratory comparison to evaluate the effectiveness of $^1$H NMR metabolomics to generate comparable data sets from environmentally derived samples. WAPT graphs and tables are used in all examples.

show abstract

Section: Univariate Modulementioning

confidence: 99%

WAPT: a friendly interactive web application for performance assessment in interlaboratory comparisons

Leal,

Rocha

2024

Preprint

View full text Add to dashboard Cite

show abstract

“…Because IRLS is iterative, it is a reliable and effective technique for handling the complexity involved in calculating M-estimators. The valuable contribution in the field of M-estimators is done by many researched and renowned names are Raza et al 8 , Mukhtar et al 9 , Luo et al 10 , Anekwe & Onyeagu 11 , Noor-ul-Amin et al 12 , Khalil et al 13 , Alamgir et al 14 , Ullah et al 15 , Ali & Qadir 16 , Qadir 17 , Hampel 18 , Andrews 19 and Beaton & Tukey 20 .…”

Section: M-estimatorsmentioning

confidence: 99%

Enhancing performance in the presence of outliers with redescending M-estimators

Raza,

Talib,

Noor-ul-Amin

et al. 2024

Sci Rep

View full text Add to dashboard Cite

In real-life situations, we have to analyze the data that contains the atypical observations, and the presence of outliers has adverse effects on the performance of ordinary least square estimates. In this situation, redescedning M-estimators, proposed by Huber (1964), are used to tackle the effects of outliers to increase the efficiency of least square estimates. In this study, we introduce a redescending M-estimator designed to generate robust estimates by mitigating the influence of outlier observations, even when the tuning constant is set to low values. This innovative estimator exhibits enhanced linearity at its core and maintains continuity throughout its range. Our proposed estimator stands out for its novelty, simplicity, differentiability, and practical applicability across real-world scenarios. The results of the proposed redescedning M-estimators are compared with existing robust estimators using an extensive simulation study. Two examples based on real-life data are also added to validate the performance of the suggested function. The formulated redescedning M-estimator produced efficient results as compared to all the considered redescedning M-estimators.

show abstract

“…For comparison, the parameters estimates of the Mean Square Error (MSE) and the absolute bias (BIAS) of the [7], [4] and [1] and the [3] redescending M-estimators are computed.…”

Section: Simulation Designmentioning

confidence: 99%

The Performance of Redescending M-Estimators when Outliers are in Two Dimensional Space

Ebele

Iheanyi

2022

EJMS

View full text Add to dashboard Cite

M-estimators are robust estimators that give less weight to the observations that are outliers while redescending M-estimators are those estimators that are built such that extreme outliers are completely rejected. In this paper, redescending M-estimators are compared using both the Monte Carlo simulation method and the real life data to ascertain the method that is more efficient and robust when outliers are in both x and y directions. The results from the simulation study and the real life data indicate that Anekwe redescending M-estimator is more efficient and robust when outliers are in both x and y directions.

show abstract

The Redescending M estimator For detection and deletion of Outliers in Regression analysis

Cited by 4 publications

References 12 publications

WAPT: a friendly interactive web application for performance assessment in interlaboratory comparisons

WAPT: a friendly interactive web application for performance assessment in interlaboratory comparisons

Enhancing performance in the presence of outliers with redescending M-estimators

The Performance of Redescending M-Estimators when Outliers are in Two Dimensional Space

Contact Info

Product

Resources

About