Discrete Dynamics in Nature and Society

Discrete Dynamics in Nature and Society / 2020 / Article

Research Article | Open Access

Volume 2020 |Article ID 4690562 | https://doi.org/10.1155/2020/4690562

Jibo Wu, Yong Li, "New Restricted Liu Estimator in a Partially Linear Model", Discrete Dynamics in Nature and Society, vol. 2020, Article ID 4690562, 6 pages, 2020. https://doi.org/10.1155/2020/4690562

New Restricted Liu Estimator in a Partially Linear Model

Academic Editor: Filippo Cacace
Received24 Jun 2020
Accepted27 Jul 2020
Published18 Aug 2020

Abstract

In this paper, we introduce a new restricted Liu estimator in a partially linear model when addition linear constraints are assumed to hold. We also consider the asymptotic normality of the new estimator. Finally, a numerical example and a simulation study are listed to illustrate the performance of the new estimator.

1. Introduction

Consider the following partially linear model:where shows a scalar dependent variable, , denotes explanatory variables, and denotes a vector of unknown parameters, and in this paper, we suppose that be an unknown smooth function, show values of an extra univariate variable such as the time at which the observation is made, and be an independent random error with and .

The statisticians have mainly discussed how to estimate the parametric component of , and many methods have been proposed to estimate such as methods by Akdeniz and Duran [1], Akdeniz et al. [2], Duran and Akdeniz [3], Akdeniz et al. [4], Akdeniz et al. [5], Heckman [6], Liang [7], Liu et al. [8], Speckman [9], Wu [10, 11], Wu and Asar [12], Yatchew [13], Yang et al. [14], and Yang and Li [15].

As we all know, in regression analysis, the presence of multicollinearity among regressor variables can lead to highly unstable least squares estimates of the regression parameters. To deal with this problem, some biased estimating methods have been proposed, such as Hoerl and Kennard [16], Liu [17], and Xu and Yang [18, 19].

The theory and approach on biased estimation were mainly with the linear regression model. How to obtain biased estimator in partially linear models is also important and interesting. In literature studies, some biased estimators have been proposed to estimate in partially linear models. Hu [20] introduced a ridge estimator by the parametric component . Liu et al. [8] introduced a PCR estimator in partially linear models. For more references, one can refer to Roozbeh [21], Roozbeh et al. [22], Akdeniz and Roozbeh [23], Roozbeh et al. [24], Roozbeh and Hanzah [25], and Wei and Wang [26].

In practice, we may find that the unknown parameter may have some restrictions, such as linear restrictions, stochastic linear restrictions, and other restrictions. Many authors have studied these linear restrictions; namely, Akdeniz and Tabakan [27] introduced the restricted ridge estimator for the parametric component and Akdeniz and Duran [1] proposed a restricted Liu estimator in a partially linear model.

As we all know, there are few papers discussing the asymptotic normality of the estimator of . In this paper, we continue this work in this aspect. First, we will introduce a new restricted Liu estimator for the parametric component when some addition linear restrictions are assumed to hold, and this estimator is different form the Liu-type estimator which Akdeniz and Duran [1] proposed. Second, we will study the asymptotic normality of the new estimator.

The rest of the paper is organized as follows. In Section 2, we propose a new restricted Liu estimator for the linear parametric component to deal with multicollinearity, and we will discuss the asymptotic normality of the new estimator in Section 3. A numerical example is given to show the performance of the proposed estimator in Section 4. A simulation study is given to show the performance of the proposed estimator in Section 5, and some conclusion remarks are listed in Section 6.

2. New Restricted Liu Estimator

In this section, based on the profile least squares estimator, we will present a new restricted Liu estimator for the unknown parameter .

To present the new estimator, we first give some assumptions (Gao [28]).

Assumption 1. There exist bounded functions over , , so thatwhere denotes real vectors which fulfill

Assumption 2. Let the functions and have the Lipschitz condition of order 1 on , . The Lipschitz condition is that with any function , there exists a constant , with any in , so that .

Assumption 3. The positive weight function fulfills the following:(i)(ii)(iii)Here, denotes the indicator function, appeases , and appeases .

2.1. Profile Least Squares Estimator

In this paper, we let satisfy partially linear model (1). When we obtain an estimator of , then we can obtain an estimator of , which is given as follows:

And we also suppose that the positive weight functions satisfy Assumption 3.

By (1) and (4), we obtain the following:where , , , and with .

To estimate , one can usually use the profile least squares method to get the profile least squares (PLS) estimator:

2.2. New Restricted Liu Estimator

In order to deal with multicollinearity, Akdeniz and Duran [1] proposed the Liu estimator (LE), which is denoted as follows:

Now, we consider the following linear restrictions:where denotes a known matrix with rank and shows a known vector. For models (1) and (8), restricted profile least squares (RPLS) estimator [1] is denoted by

We are now ready to introduce a new restricted Liu estimator, which is obtained by combining the Liu estimator proposed by Akdeniz and Duran [1] and the restricted profile least squares (RPLS) estimator as follows:where is the Liu estimator. By (10), we can see that the new restricted Liu estimator is different from the restricted Liu estimator proposed by Akdeniz and Duran [1], and so this Liu estimator is a new restricted estimator.

From the definition of , it is easy to see that .

In this paper, we mainly discuss how to estimate . For the estimator of , we will discuss it in the next study.

3. Properties of the New Estimator

To obtain the properties of the new estimator, we first list some lemmas.

Lemma 1. If Assumptions 13 hold, we have

Proof. See Gao et al.’s study [28].

Lemma 2. Under Assumptions 13, PLS estimator is an asymptotic normality of , i.e.,where denotes convergence in distribution.

Proof. See Gao et al.’s study [28].

Theorem 1. Under Assumptions 13, Liu estimator is an asymptotic normality of , i.e.,

Proof. By (7), we haveThen,By Lemma 1, it is easy to derive thatwhere shows convergence in probability and denotes the infinitesimal of higher order. Thus, by Lemma 2 and (15)–(18), we obtainThen, by the Slutsky theorem and (15) and (19), we obtainNow, we present the asymptotic normality of the new restricted Liu estimator.

Theorem 2. Under Assumptions 13, the new restricted Liu estimator is an asymptotic normality of , i.e.,where .

Proof. By (10), we obtainBy Lemma 1 and (22), we obtainThus by (23), we havewhere . Thus, by (22)–(24) and the Slutsky theorem, we derive

Remark 1. By the asymptotic covariance matrices of and , we note that is a positive definite matrix. That is, to mean if the linear restrictions (8) are assumed to hold, the new restricted Liu estimator is more efficient than the Liu estimator , and this agrees with the practice; thus, the new restricted Liu estimator contains more information for than the Liu estimator.

4. Numerical Example

In this section, we will use a numerical example to illustrate the performance of the new estimator. We consider the hedonic prices of housing attributes. The partially linear model was estimated by Ho [29] using semiparametric least squares. The data consist of 92 detached homes sold during 1987 in the Ottowa area [30].

The specification of the partially linear model is

The matrix has eigenvalues and the condition number of is 250.069, which is large. The Gaussian kernel functionis used when obtaining the matrices and is taken to be 0.5. We use the method presented by Wu and Asar [12] to estimate . In these data, we choose , where

And we use Liu [17] to estimate . Then, we obtain

By the numerical example, we can see that the new estimator has smaller MSE than other estimators.

5. Monte Carlo Simulation Study

5.1. Design of the Simulation

In this section, we use a Monte Carlo simulation to discuss the performances of the estimators PLS, LE, and RPLS and the new estimator in the sense of MSE criterion. Although the purpose of this paper is to compare the estimators when there is multicollinearity, we use the following equation to obtain the explanatory variables having different degrees of multicollinearity:where show independent standard normal pseudo-random numbers. We form the data matrix as . We discuss three different values of the degree of correlation corresponding to , and 0.999.

In order to obtain the observations of the dependent variable, we first use the following Doppler function:where , , and the sample size varies between 30, 50, 100, 200, and 400 in this simulation study. After estimating the function , we use equation (1) to obtain the dependent variables. Finally, we estimate the function again using equation (4).

The Gaussian kernel functionis used when obtaining the matrices and is taken to be 0.05. In this paper, we study and consider the following restriction:Moreover, we also consider .

5.2. Results of the Simulation

The results of the simulation are presented in Tables 15. From Tables 15, we see that an increase in the degree of correlation makes an increase in the simulated MSE values of the estimators in all cases.



0.118420.058370.118130.05814
0.165190.081430.164740.08098
0.193990.095630.193470.09501



0.164200.083960.164000.08346
0.229110.117120.229100.11615
0.260960.137540.269500.13622



0.069980.041280.069880.04210
0.097620.058830.097450.05871
0.114600.069090.114420.06891



0.037370.022930.037330.02291
0.052130.031980.052060.03194
0.061220.037560.061130.03750



0.020540.012710.020490.01270
0.028590.017720.028580.01771
0.033580.020810.033560.02080

Moreover, the asymptotic behaviours of the estimators are satisfied in all of the cases, namely, an increase in the sample size makes a decrease in the MSE values. And the new estimator performs best compared to other estimators in all cases.

6. Conclusions

In this paper, we propose a new restricted Liu estimator when some additional linear restrictions are assumed to hold on the linear parametric component. And we study the properties of the proposed estimator. Finally, a data example and a simulation study are given to show the performance of these estimators.

Data Availability

The data used to support the findings of this study are available from the corresponding author upon request.

Conflicts of Interest

The authors declare that they have no conflicts of interest.

Acknowledgments

This work was sponsored by the National Natural Science Foundation of China (Grant no. 11501072), the Natural Science Foundation of Chongqing (Grant no. cstc2019jcyj-msxmX0379), and the Scientific Technological Research Program of Chongqing Municipal Education Commission (Grant no. KJQN201901347).

References

  1. F. Akdeniz and E. A. Duran, “Liu-type estimator in semiparametric regression models,” Journal of Statistical Computation and Simulation, vol. 80, no. 8, pp. 853–871, 2010. View at: Publisher Site | Google Scholar
  2. D. E. Akdeniz, F. Akdeniz, and H. Hu, “Efficiency of a Liu-type estimator in semiparametric regression models,” Journal of Computational and Applied Mathematics, vol. 235, no. 5, pp. 1418–1428, 2011. View at: Publisher Site | Google Scholar
  3. D. E. Akdeniz and F. Akdeniz, “New difference-based estimator of parameters in semiparametric regressionmodels,” Journal of Statistical Computation and Simulation, vol. 83, pp. 810–824, 2013. View at: Publisher Site | Google Scholar
  4. F. Akdeniz, E. Akdeniz Duran, M. Roozbeh, and M. Arashi, “Efficiency of the generalized difference-based Liu estimators in semiparametric regression models with correlated errors,” Journal of Statistical Computation and Simulation, vol. 85, no. 1, pp. 147–165, 2015. View at: Publisher Site | Google Scholar
  5. E. Akdeniz, F. Akdeniz, and M. Roozbeh, “A new difference-based weighted mixed Liu estimator in partially linear models,” Statistics, vol. 52, no. 6, pp. 1309–1327, 2018. View at: Publisher Site | Google Scholar
  6. N. E. Heckman, “Spline smoothing in a partly linear model,” Journal of the Royal Statistical Society: Series B (Methodological), vol. 48, no. 2, pp. 244–248, 1986. View at: Publisher Site | Google Scholar
  7. H. Liang, “Estimation in partially linear models and numerical comparisons,” Computational Statistics & Data Analysis, vol. 50, no. 3, pp. 675–687, 2006. View at: Publisher Site | Google Scholar
  8. C. Liu, S. Guo, and C. Wei, “Principal components regression estimator of the parameters in partially linear models,” Journal of Statistical Computation and Simulation, vol. 86, no. 15, pp. 3127–3133, 2016. View at: Publisher Site | Google Scholar
  9. P. Speckman, “Kernel smoothing in partial linear models,” Journal of the Royal Statistical Society: Series B (Methodological), vol. 50, no. 3, pp. 413–437, 1998. View at: Publisher Site | Google Scholar
  10. J. Wu, “Restricted difference-based Liu estimator in partially linear model,” Journal of Computational and Applied Mathematics, vol. 300, pp. 97–102, 2016. View at: Publisher Site | Google Scholar
  11. J. Wu, “Improved Liu-type estimator in partial linear model,” International Journal of Computer Mathematics, vol. 93, no. 3, pp. 498–510, 2016. View at: Publisher Site | Google Scholar
  12. J. Wu and Y. Asar, “A weighted stochastic restricted ridge estimator in partially linear model,” Communications in Statistics - Theory and Methods, vol. 46, no. 18, pp. 9274–9283, 2017. View at: Publisher Site | Google Scholar
  13. A. Yatchew, “An elementary estimator of the partial linear model,” Economics Letters, vol. 57, no. 2, pp. 135–143, 1997. View at: Publisher Site | Google Scholar
  14. H. Yang, J. Lv, and C. Guo, “Robust estimation and variable selection for varying-coefficient single-index models based on modal regression,” Communications in Statistics-Theory and Methods, vol. 45, no. 14, pp. 4048–4067, 2016. View at: Publisher Site | Google Scholar
  15. H. Yang and N. Li, “WLAD-LASSO method for robust estimation and variable selection in partially linear models,” Communications in Statistics - Theory and Methods, vol. 47, no. 20, pp. 4958–4976, 2018. View at: Publisher Site | Google Scholar
  16. A. E. Hoerl and R. W. Kennard, “Ridge regression: Biased estimation for nonorthogonal problems,” Technometries, vol. 12, no. 1, pp. 55–67, 1970. View at: Publisher Site | Google Scholar
  17. K. Liu, “A new class of biased estimate in linear regression,” Communications in Statistics-Theory and Methods, vol. 22, no. 2, pp. 393–402, 1993. View at: Publisher Site | Google Scholar
  18. J. Xu and H. Yang, “More on the bias and variance comparisons of the restricted almost unbiased estimators,” Communications in Statistics-Theory and Methods, vol. 40, no. 22, pp. 4053–4064, 2011. View at: Publisher Site | Google Scholar
  19. J. Xu and H. Yang, “On the restricted almost unbiased estimators in linear regression,” Journal of Applied Statistics, vol. 38, no. 3, pp. 605–617, 2011. View at: Publisher Site | Google Scholar
  20. H. Hu, “Ridge estimation of a semiparametric regression model,” Journal of Computational and Applied Mathematics, vol. 176, no. 1, pp. 215–222, 2015. View at: Publisher Site | Google Scholar
  21. M. Roozbeh, “Optimal QR-based estimation in partially linear regression models with correlated errors using GCV criterion,” Computational Statistics & Data Analysis, vol. 117, pp. 45–61, 2018. View at: Publisher Site | Google Scholar
  22. M. Roozbeh, S. Babaie-Kafaki, and A. Naeimi Sadigh, “A heuristic approach to combat multicollinearity in least trimmed squares regression analysis,” Applied Mathematical Modelling, vol. 57, pp. 105–120, 2018. View at: Publisher Site | Google Scholar
  23. F. Akdeniz and M. Roozbeh, “Generalized difference-based weighted mixed almost unbiased ridge estimator in partially linear models,” Statistical Papers, vol. 60, no. 5, pp. 1717–1739, 2019. View at: Publisher Site | Google Scholar
  24. M. Roozbeh, M. Arashi, and N. A. Hamzah, “Generalized cross-validation for simultaneous optimization of tuning parameters in ridge regression,” Iranian Journal of Science and Technology, Transactions A: Science, vol. 44, no. 2, pp. 473–485, 2020. View at: Publisher Site | Google Scholar
  25. M. Roozbeh and N. A. Hamzah, “Uncertain stochastic ridge estimation in partially linear regression models with elliptically distributed errors,” Statistics, vol. 54, no. 3, pp. 494–523, 2020. View at: Publisher Site | Google Scholar
  26. C. Wei and X. Wang, “Liu-type estimator in semiparametric partially linear additive models,” Journal of Nonparametric Statistics, vol. 28, no. 3, pp. 459–468, 2016. View at: Publisher Site | Google Scholar
  27. F. K. Akdenz and G. Tabakan, “Restricted ridge estimators of the parameters in semiparametric regression model,” Communications in Statistics-Theory and Methods, vol. 38, no. 11, pp. 1852–1869, 2009. View at: Publisher Site | Google Scholar
  28. J. T. Gao, S. Y. Hong, and H. Liang, “Convergence rates of a class of estimates in partly linear models,” Acta Mathematica Sinica, vol. 38, pp. 658–669, 1995. View at: Google Scholar
  29. M. Ho, “Essays on the housing market,” University of Toronto, Toronto, Canada, 1995, Ph. D. diss. View at: Google Scholar
  30. A. Yatchew, Semiparametric Regression for the Applied Econometrician, Cambridge University Press, Cambridge, United Kingdom, 2003.

Copyright © 2020 Jibo Wu and Yong Li. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.


More related articles

 PDF Download Citation Citation
 Download other formatsMore
 Order printed copiesOrder
Views145
Downloads196
Citations

Related articles

Article of the Year Award: Outstanding research contributions of 2020, as selected by our Chief Editors. Read the winning articles.