ArticlePDF Available

Proposing Robust IRWs Technique to Estimate Segmented Regression Model for the Bed load Transport of Tigris River with Change Point of Water Discharge Amount at Baghdad City

Authors:

Abstract

Segmented regression consists of several sections separated by different points of membership, showing the heterogeneity arising from the process of separating the segments within the research sample. This research is concerned with estimating the location of the change point between segments and estimating model parameters, and proposing a robust estimation method and compare it with some other methods that used in the segmented regression. One of the traditional methods (Muggeo method) has been used to find the maximum likelihood estimator in an iterative approach for the model and the change point as well. Moreover, robust estimation method (IRWm method) has used which depends on the use of the robust M-estimator technique in segmentation idea and using the Tukey weight function. Our contribution to this research lies in the suggestion to use the S-estimator technique and using the Tukey weight function, to obtain a robust method against cases of violation of the normal distribution condition for random errors or the effect of outliers, and this method will be called IRWs. The aforementioned methods have been applied to a real data set related to the bed-load of Tigris River/ Baghdad city as a response variable and the amount of water discharge as an explanatory variable. The results of the comparison showed the superiority of the proposed method.
Journal of Economics and Administrative Sciences
Vol.26 (NO. 121) 2020, pp. 415-427

Available online at http://jeasiq.uobaghdad.edu.iq
IRWs







)dromar72@coadec.uobaghdad.edu.iq(
)mohammad.stat.145@gmail.com(
Published:11/3/2020 Accepted:3/5/2020 Received :August / 2020
NC 4.0)-(CC BY NonCommercial 4.0 International-Attribution
 



         Muggeo    
     IRWm
    M-estimator       Tukey
      S-estimator   Tukey
IRWs


         M
S

Journal of Economics and Administrative Sciences
Vol.26 (NO. 121) 2020, pp. 415-427

-1Introduction
     Segmented linear regression model   


join point
(change point)(break points)
   [10]broken line model     ([16] joinpoint regression model)[17]piecewise linear model






      
     ([11][12]Quandt,1958,1960)
([9]Hudson,1966), ([6]Feder,1975), ([7]Gallant and Fuller,1973)   
 
      [10]Muggeo,2003)  
(Linear reparametrization technique)

    


([17]F. Zhang and Q. Li,2017)
   robust rank-based estimator bent linear regression    
       Muggeo,2003)     rank dispersion
([4]O.A.Ali, M.A.Abbas,2019)
     (M-Estimator)    
HuberTukeyHampel(Muggeo)
              
 (Tukey)  [1]Sukru A., Birdal S.,2020
[9] Hudson 1966 [10] Muggeo2003  (MML) 
  (two-segmented linear regression model)
 


               
(S-estimator)(Muggeo)(S-estimator)   MM-scales  
(M-estimator)
Journal of Economics and Administrative Sciences
Vol.26 (NO. 121) 2020, pp. 415-427

2
2-1
      
            
[8:pp.569]



(





(1)
 

0


2-2Estimation methods1-2-2(Muggeo)
    (maximum likelihood)      
(ML)

([10]Vito M. R. Muggeo,2003)

 
[10:pp3057]
       
 
 
(3)
 

reparameterization


x

(Muggeo)

 (Taylor expansion)    
  
 
[17:pp.6]
Journal of Economics and Administrative Sciences
Vol.26 (NO. 121) 2020, pp. 415-427


(5)(6)
 (5)




 



    (ML)

[10:pp.3059-3061]



(10) 

(9)  (ML)



(
(9)3
24
4s

4
ML
sth
2-2-2MRobust IRWm-estimator
  (IRWm-estimator)        ([4]O.A.Ali, M.A.Abbas,2019)(M-estimator)
(Muggeo)
 (loss function
            [4:pp.393-394]


[4:pp.394-395]
  


     
0.01
 




       M
(9)(Muggeo) a 

 
(9)Muggeo

Journal of Economics and Administrative Sciences
Vol.26 (NO. 121) 2020, pp. 415-427

b  c  




M
[5:pp56]


 






 



(14)


 
(15) (WLS)
[5:pp.57]

n n
d ac[17:pp.7] 







(11)(Muggeo)


(Muggeo)
[10:pp.3061]     
sth

(IRWm-estimator)
p

(Tukey) ([4]O.A.Ali, M.A.Abbas,2019)
(Tukey bisquare)[14:pp.6]




c = 4.685
Journal of Economics and Administrative Sciences
Vol.26 (NO. 121) 2020, pp. 415-427

3-2-2    (S-estimator)   

  S       MM-scales 
[13]Rousseeuw and Yohai,1984        M
         weaknesses   [15:pp.354]
         (Muggeo)
M(2-2-2)

S
(IRWs-estimator)

  


    
0.01
 




      S
(9)(Muggeo) a
 

 
(9)Muggeo
 b
 c
  




   
S[15:pp. 354]


 





 

 


(Tukey)



 

 
(19)(2-2-2)
s(WLS)

[15:pp. 355]

Journal of Economics and Administrative Sciences
Vol.26 (NO. 121) 2020, pp. 415-427

n n
d
ac[17:pp.7] 







(11)(Muggeo)


            (Muggeo)
[10:pp.3061]     
sth

(IRWs-estimator)
2-3The standard error for the change point
             Wald    Wald    
 
[17:pp.8]


 
var(.)cov(. , ,)


 converges(Muggeo) 

[10:pp.3061]



3The Application
3-1Real Data Description
 (Sediment transport)  


      
[2:pp.7-8]
8

([3]Ammar A. Ali, Nadhir A. Al-Ansari, Qusay Al-Suhail and
Sven Knutsson,2017)   16 
(1)
[3:pp.67] CS6-1
8(1)(2)
Journal of Economics and Administrative Sciences
Vol.26 (NO. 121) 2020, pp. 415-427

:(1)
[3:p p.60]
(1)CS6-1 
(van Rijn 1984)
1
1000
900
800
700
600
500
400
300
X
4.011
3.122
2.529
2.157
2.003
1.866
1.954
2.251
Y
[3]
1
 XDischarge   Y bed load  
Journal of Economics and Administrative Sciences
Vol.26 (NO. 121) 2020, pp. 415-427

(2)    CS6-1   
(van Rijn 1984)
3-2Results Discussion

          
(2)
(3)(S.E.( ))
   (MSE)       
(2)
(2)
CS6-1 
Parameters of the
segmented linear
regression model
Methods
Muggeo
IRWm-
estimator
IRWs-estimator
Tukey Bisquare
Tukey Bisquare
737.0557
737.4091
733.4267
2.1157
2.1049
1.8560
-0.00014
-0.00012
0.00025

-3.4483
-3.4475
-3.3990
0.00741
0.00741
0.00741
S.E.( )
34.151
33.203
0.124
MSE
0.02743
0.02522
7.6698e-8
Journal of Economics and Administrative Sciences
Vol.26 (NO. 121) 2020, pp. 415-427

Muggeo
IRWm-estimator(Tukey Bisquare)
IRWs-estimator(Tukey Bisquare)
(3)CS6-1 

 (2)    (MSE
    S.E.( )  (IRWs-estimator)   (Tukey bisquare)(IRWm-estimator)(Tukey bisquare)
4Conclusions and Recommendations
4-1Conclusions

(IRWs-estimator)(Tukey)
    ML (Muggeo)  (IRWm-estimator) 
(MSE)S.E.( )     (IRWs-estimator)   (Tukey) 
              
         
  (IRWs-estimator) (Tukey)

.
Journal of Economics and Administrative Sciences
Vol.26 (NO. 121) 2020, pp. 415-427

4-2Recommendations


(IRWs-estimator)
 3-4Future Studies
     
              




              

Acknowledgments
We introduce our thanks to the editors and the reviewers of (JEAS) journal for
their efforts in enriching the research with their valuable comments. Thanks also
to Dr. Ammar Adel Ali for providing us with data on the hydrology of the Tigris
River at Baghdad city. References
1. Acitas,S., Senoglu,B.,(2020). Robust change point estimation in two-phase
linear regression models: An application to metabolic pathway data, Journal of
Computational and Applied Mathematics,Vol.363,pp. 337349.
2. Ali A. A .(2016) "Three Dimensional Hydro- Morphological of Tigris River". A
thesis submitted to Lulea University of Technology. Sweden .
3. Ali, A.A., Al-Ansari N.A., Al-Suhail, Q., and Knutsson S., (2017). Spatial
Measurement of Bedload Transport in Tigris River. Journal of Earth Sciences and
Geotechnical Engineering, vol.7, no. 4, 55-75.
4. Ali,O.A., Abbas,M.A.,(2019),New Robust Estimator Of Change Point In
Segmented Regression Model For Bed-Load Of Rivers, Journal of Mechanics of
Continua and Mathematical Sciences, Vol.14, No.6, pp.384-402.
5. Almetwally,E.M., Almongy,H.M.,(2018), Comparison Between M-estimation, S-
estimation, And MM Estimation Methods of Robust Estimation with Application
and Simulation, International Journal of Mathematical Archive, 55-63 .
6. Feder,P.I.,(1975),The log likelihood Ratio In Segmented Regression, Annals of
Statistics, pp. 84-97.
7. Gallant,A. R., Fulle, W.A. ,(1973), Fitting Segmented Polynomial Regression
Models Whose Join Points have to be Estimated, Journal of the American Statistical
Association, Vol. 68, No. 341, pp. 144-147.
8. Greene, M.E., Rolfson, O.,Garellick, G.,Gordon, M. and Nemes, S., (2014),
Improved statistical analysis of pre- and post-treatment patient-reported
outcome measures (PROMs) the applicability of piecewise linear regression
splines, Quality of Life Research, Vol.24,Issue3,pp.567573.
Journal of Economics and Administrative Sciences
Vol.26 (NO. 121) 2020, pp. 415-427

9. Hudson, D.J., (1966), Fitting Segmented Curves Whose Join Points Have to be
Estimated, Journal of the American Statistical Association, 61:316, 1097-1129.
10. Muggeo, V. M.R., (2003),Estimating regression models with unknown
breakpoints,Statistics in Medicine 22, 30553071.
11. Quandt, R.E., (1958),The estimation of the parameters of a linear regression
system obeying two separate regimes. J. Amer. Statist. Assoc,53,873880.
12. Quandt, R.E., (1960), Tests of the Hypothesis That a Linear Regression System
Obeys Two Separate Regimes, J. Amer. Statist. Assoc,55:290, 324-330.
13. Rousseeuw, P. J, Yohai, V. J.,(1984).Robust Regression by Mean of
SEstimators,Robust and Nonlinear Time Series Analysis, New York, ,256-274, doi:
10.1007/978-1-4615-7821-5-15.
14. Salini, S., Cerioli, A., Laurini, F. and Riani, M.,(2015), Reliable Robust
Regression Diagnostics, International Statistical Review, 0, 0, 129.
15. Susanti, Y., Pratiwi, H., Sulistijowati, H.S. and Liana, T.,(2014).M estimation, S
estimation, and MM Estimation in Robust Regression. Int. J. Pure Appl. Math. Vol.
91 No. 3,pp. 349-360.
16. Yu, B., Barrett, M. J., Kim, H.-J. and Feuer, E. J. (2007). Estimating joinpoints in
continuous time scale for multiple change-point models. Computational Statistics &
Data Analysis51, 24202427.
17. Zhang, F., Li, Q.,(2017), Robust bent line regression, Journal of Statistical
Planning and Inference.
Journal of Economics and Administrative Sciences
Vol.26 (NO. 121) 2020, pp. 415-427

Proposing Robust IRWs Technique to Estimate Segmented Regression Model
for the Bed load Transport of Tigris River with Change Point of Water
Discharge Amount at Baghdad City
Mohammed Ahmed Abbas
Omar Abdulmohsin Ali
Researcher. Sunni Endowment
Diwan, Department of Religious
Education and Islamic Studies,
Baghdad, Iraq
Assist. Prof. Department of
Statistics, College of Management
and Economics, Baghdad
University, Baghdad, Iraq
(mohammad.stat.145@gmail.com)
(dromar72@coadec.uobaghdad.edu.iq)
Published:11/3/2020 Accepted:3/5/2020 Received :August / 2020
NonCommercial 4.0 -Creative Commons AttributionThis work is licensed under a NC 4.0)-(CC BY International
Abstract
Segmented regression consists of several sections separated by different
points of membership, showing the heterogeneity arising from the process of
separating the segments within the research sample. This research is concerned
with estimating the location of the change point between segments and estimating
model parameters, and proposing a robust estimation method and compare it
with some other methods that used in the segmented regression. One of the
traditional methods (Muggeo method) has been used to find the maximum
likelihood estimator in an iterative approach for the model and the change point
as well. Moreover, robust estimation method (IRW method) has used which
depends on the use of the robust M-estimator technique in segmentation idea and
using the Tukey weight function. The research’s contribution lies in the
suggestion to use the S-estimator technique and using the Tukey weight function,
to obtain a robust method against cases of violation of the normal distribution
condition for random errors or the effect of outliers, and this method will be
called IRWs. The mentioned methods have been applied to a real data set related
to the bed-load of Tigris River/ Baghdad city as a response variable and the
amount of water discharge as an explanatory variable. The results of the
comparison showed the superiority of the proposed method.
Keywords: segmented linear regression, change point, reparameterization,
M-estimator, S-estimator, bed-load transport, hydrology of water bodies.
Preprint
Full-text available
In this research, the segmented regression model with multiple change points will be dealt with. the maximum likelihood estimator (MLE) and the two methods of robust repetitive weights (IRWm), (IRWs) will be used to estimate the model parameters and change points and then compare between these methods to choose the best method between them. A simulation process will be created with several scenarios, with different sample sizes and contamination rates (Outliers values) (15%, 10%, 5%, 0%) to Procedure the comparison process and find the best method for estimation. The simulation results, after completing the comparison process using the comparison standard, mean squared error (MSE), showed the efficiency of the Iterative Weights method (IRWm) when there were contamination rates in the data, and the efficiency of the maximum likelihood method (MLE) when the data did not contain any contamination.
Article
Full-text available
Segmentation has vital employment in regression analysis where data have some change point. Traditional estimation methods such as Hudson, D.J.;(1966) and Muggeo, V. M., (2003) have been reviewed. But these methods do not take into account robustness in the presence of outliers values. However, third method was used as rank-based method, where the analysis will be devoted to the ranks of data instead of the data themselves. Our contribution in this paper is to use M-estimator methodology with three distinct weight functions (Huber, Tukey, and Hampel) which has been combined with Muggeo version approach to gain more robustness, Thus we get robust estimates from the change point and regression parameters simultaneously. We call our new estimator as robust Iterative Rewrighted M-estimator:IRWm-method with respect to its own weight function. Our primary interest is to estimate the change point that joins the segments of regression curve, and our secondary interest is to estimate the parameters of segmented regression model. The real data set were used which concerned to bed-loaded transport as dependent variable (y) and discharge explanatory variable (x). The comparison has been conducted by using several criteria to select the most appropriate method for estimating the change point and the regression parameters. The superior results were marked for IRWm-estimator with respect to Tukey weight function.
Article
Full-text available
In regression analysis the use of ordinary least squares, (OLS) method would not be appropriate in solving problem containing outlier or extreme observations. Therefore, we need a method of robust estimation where the value of the estimation is not much affected with these outlier or extreme observations. In this paper, six methods of estimation will be compared in order to reach the best estimation, and these methods are M.Humpel estimation method, M.Bisquare estimation method, M.Huber estimation method, S-estimation method, MM(S)-estimation method, and MM estimation method in robust regression to determine a regression model. We find that, the best three method, through this study, are M-estimation method, MM(S)-estimation method and MM estimation method. Since M-estimation method is an extension of the maximum likelihood method, while MM estimation method is the development of M-estimation method and MM(S) estimation method is the development of S-estimation method. Robust regression methods can considerably improve estimation precision, but should not be applied automatically instead of the classical methods.
Article
Full-text available
Using Helley-Smith sampler, 288 bed load samples were collected from 16 cross sections along 18 km reach length of Tigris River within Baghdad. The spatial distribution of sampling along the reach took into consideration the variance of river topography where 7 meanders, 2 islands and several bank depositions characterize the geometry of the river. The implemented regulation schemes on Tigris River have reduced 44% of water discharges compared to previous period. The spatial variance in topography was effectively scattering the results of the applied twenty bed load formulas. The study results indicated that the complicated geometry of the river reach makes finding a unique representative bed load formula along the study reach rather difficult, and there is no grantee to have good agreement with measurements in the irregular cross sections (meanders, sand bars, etc.). The closest bed load prediction formulas were van Rijn1984. The annual transported quantities of bed load were estimated to be 30 thousand tons (minimum) in 2009 and 50 thousand tons (maximum) in 2013.
Article
Full-text available
Motivated by the requirement of controlling the number of false discoveries that arises in several application fields, we study the behaviour of diagnostic procedures obtained from popular high-breakdown regression estimators when no outlier is present in the data. We find that the empirical error rates for many of the available techniques are surprisingly far from the prescribed nominal level. Therefore, we propose a simulation-based approach to correct the liberal diagnostics and reach reliable inferences. We provide evidence that our approach performs well in a wide range of settings of practical interest and for a variety of robust regression techniques, thus showing general appeal. We also evaluate the loss of power that can be expected from our corrections under different contamination schemes and show that this loss is often not dramatic. Finally, we detail some possible extensions that may further enhance the applicability of the method.
Article
Full-text available
In regression analysis the use of least squares method would not be appropriate in solving problems containing outlier or extreme observations. So we need a parameter estimation method which is robust where the value of the estimation is not much affected by small changes in the data. We present M estimation, S estimation and MM estimation in robust regression to determine a regression models. M estimation is an extension of the maximum likelihood method and is a robust estimation, while S estimation and MM estimation are developments of the M estimation method. The algorithm of these methods is presented and we apply them on the maize production data.
Article
Full-text available
Purpose: Patient-reported health-related quality-of-life (HRQoL) measures such as the EuroQol 5 dimension (EQ-5D) index are commonplace when assessing healthcare providers or efficiency of medical techniques. HRQoL measures are generally bounded, and the magnitude of possible improvement depends on the pre-treatment HRQoL value. This paper aimed to assess and illustrated the possibility of modelling the relationship between pre- and post-treatment HRQoL measures with piecewise linear splines. Methods: The method was illustrated using a longitudinal dataset of 36,625 patients with one EQ-5D index before and one a year after total hip arthroplasty. We considered four models: intercept only model, single line regression, and segmented regression with 1 and 2 change points. The post-operative EQ-5D index served as the outcome, while the preoperative EQ-5D index was the predictor. Results: We found that a two-line regression best described the data with the lines meeting at 0.159 on the preoperative EQ-5D index scale. In the low preoperative group (with an initial preoperative index from -0.594 to 0.159), the predicted post-operative scores ranged from 0.368 to 0.765, with post-operative scores increasing 0.528 points for each unit in the preoperative score. In the high preoperative group (initial range from 0.159 to 1), the predicted post-operative scores ranged from 0.765 to 0.855, increasing 0.106 points for each unit in the preoperative score. Conclusions: Piecewise linear regression is a straightforward approach to analyse baseline and follow-up HRQoL measures such as the EQ-5D index. It can provide a reasonable approximation of the shape of the underlying relationship where the threshold and slopes prove informative and meaningful.
Article
Full-text available
There are at least two reasons why robust regression techniques are useful tools in robust time series analysis. First of all, one often wants to estimate autoregressive parameters in a robust way, and secondly, one sometimes has to fit a linear or nonlinear trend to a time series. In this paper we shall develop a class of methods for robust regression, and briefly comment on their use in time series. These new estimators are introduced because of their invulnerability to large fractions of contaminated data. We propose to call them “S-estimators” because they are based on estimators of scale.
Article
In this study, we develop robust versions of the change point estimation methods given by Hudson (1966) and Muggeo (2003) in the two-phase linear regression model. We use a modified maximum likelihood (MML) methodology originated by Tiku (1967, 1968) when the error terms of a two-phase linear regression model are independently and identically distributed as long-tailed symmetric. Proposed estimators are shown to be more efficient and robust using the Monte-Carlo simulation. Julious’s (Julious, 2001) metabolic pathway data is analyzed in the application part of the study. It is shown that for this data using a LS estimator is inappropriate since there is an outlying observation. We therefore use proposed robust estimators instead of LS estimators and obtain more reliable results.
Article
The study considers the problem of finding the least squares estimates for the unknown parameters of a regression model which consists of grafted polynomial submodels. The abscissae of the join points are a subset of the unknown parameters. Examples are given to illustrate how continuity and differentiability conditions on the model can be used to reparameterize the model so as to allow Modified Gauss-Newton fitting. A slightly generalized version of Hartley's theorem is stated to extend the Modified Gauss-Newton method to this problem.