Missing observations: The loss in relative A-, D- and G-efficiency

 
 
 
  • Abstract
  • Keywords
  • References
  • PDF
  • Abstract


    The loss in Relative A-, D- and G-efficiency due to missing single or multiple observations is studied using cuboidal designs associated with response models. Higher losses in Relative A- and D-efficiencies are attributed to missing vertex points. The absence of one or two center points does not affect any of Relative A-, D- and G-efficiency, but when its absence is in combination with either a vertex or axial point, there is some negative effect on the design efficiency resulting in some percentage loss in Relative efficiency. The loss in relative efficiency is higher when the missing center point is in combination with missing vertex point. Losses in Relative A- and D-efficiencies are generally higher than losses in Relative G-efficiency. In fact, Relative G-efficiency is mildly affected by the missing vertex or axial point or both.


  • Keywords


    : Missing Observations; Cuboidal Design; Loss in Relative A-Efficiency; Loss in Relative D-Efficiency; Loss in Relative G-Efficiency.

  • References


      [1] T. Ahmad, Some new Second-order Response Surface Designs robust to Missing Observations,Thesis in Statistics, TheIslamia University of Bahawalpur,Thesis Abstractat Pakistan Journal of Statistics and Operations Research, (2011).

      [2] M. Akhtar, One or two missing observations in five factor Box and Behnken Design,Journal of Engg.& App. Scs.,6(1) (1987) 87-89.

      [3] M. Akhtar, Five-factor Central Composite Designs robust to a pair of missing Observations, Journal of Research (Science), Bahauddin Zakariya University, Multan, Pakistan, 12 (2) (2001) 105-115.

      [4] M. Akhtar, P. Prescott, Response Surface Designs Robust to missing observations, Communications in Statistics- Simulation, 15 (2) (1986) 245-363.

      [5] M. Akhtar, P. Prescott, A review of Response Surface Designs, Pak. Journal of Statistics, 3 (3) B (1987) 11-26.

      [6] M. Akram, Central Composite Designs Robust to three missing observations, A Ph. D Thesis in Statistics, Islamia University, Bahawalpur, (2002).

      [7] D. F. Andrews, A. M. Herzhberg, The Robustness and optimality of Response Surface Designs, Journal of Statistical Planning and Inference, 3 (1979) 249-257. https://doi.org/10.1016/0378-3758(79)90016-8.

      [8] M. A. da Silva, S. G. Gilmour, L. A.Trinca, Factorial and Response Surface Designs robust to missing observation, Computational Statistics and Data Analysis,Available online on 7 June 2016.

      [9] A. M. Herzhberg, D. F. Andrews, Some considerations in the optimal design of experiments in Non-optimal situation, Journal Roy. Statist. Soc. B38 (1976) 284-289.

      [10] R. H., Myers, D. C. Montgomery, C. M. Anderson-Cook, Response Surface Methodology: Process and Product Optimization using designed experiments, 3rd Edition, John Wiley & Sons, Inc. New Jersey, (2009).

      [11] E. A. Rady, M. M. E. Abd El-Monsef, M. M. Seyam, Relationship among several optimality criteria, interstat.statjournals.net> YEAR> articles (2009).

      [12] B. J. Smucker, W. Jensen, Z. Wu, B. Wang, Robustness of Classical andOptimal Designs to Missing Observations, Computational Statistics & Data Analysis, Available online January 2017, 17 pages, http://doi.org/10.1016/j.csda.2016.12.001.

      [13] P. Srisuradetchai, Robust response surface designs against missing observations. A Ph. D dissertation, Montana State University, Bozeman, Montana, (2015).

      [14] D. C. Whittinghill, A note on the Robustness of Box-Behnkens Designs to unavailability of Data, Metrika, 48 (1998) 49-52. https://doi.org/10.1007/PL00003971.

      [15] Y.Yakubu, A. U.Chukwu, B. T. Adebayo, A. G. Nwanzo, Effects of missing observations on predictive capability of central composite designs, International Journal on Computational Sciences & Applications, 4 (6) (2014) 1-18. https://doi.org/10.5121/ijcsa.2014.4601.


 

View

Download

Article ID: 7786
 
DOI: 10.14419/ijams.v5i2.7786




Copyright © 2012-2015 Science Publishing Corporation Inc. All rights reserved.