Çarpık ve Kirlenmiş Veriler Altında Tekrar Sayısı ve Tahminci Seçiminin İkili Yanıt Yüzeyi Optimizasyonuna Etkisi

Öz

Deney noktası içindeki tekrar sayısı ve tahminci seçimi, ikili yanıt yüzeyi optimizasyonunda çoğunlukla ayrı ayrı ele alınmaktadır. Bu çalışmada bu iki unsurun temiz, çarpık ve eklemeli olarak kirlenmiş veriler altındaki ortak etkisi, 39 senaryo, 200 Monte Carlo tekrarı ve r ∈ {3,5,8,10} için iç içe örneklemler kullanılarak değerlendirilmiştir. Ana analizde çarpıklığa göre düzeltilmiş bir konum tahmincisi sabit tutulmuş ve altı klasik ve sağlam ölçek tahmincisi karşılaştırılmıştır. Karar kalitesi tam-süreç ve temel-süreç hedeflerine göre değerlendirilmiştir. Tekrar sayısının 3’ten 8’e çıkarılması, tam-süreç için ortalama optimizasyon kaybını temiz veride %66,9, yalnızca çarpık veride %40,5 ve düşük, orta ve yüksek kirlenme olasılıklarında sırasıyla %38,7, %24,6 ve %22,5 azaltmıştır. Ortak konum–ölçek analizinde aynı geçiş, 80 birleşimin 79’unda tam-süreç, tamamında ise temel-süreç kaybını iyileştirmiştir. Hiçbir tahminci bütün koşullarda üstün değildir. Klasik standart sapma temiz veride ve kirlenmenin operasyonel sürecin parçası sayıldığı birçok durumda en iyi performansı göstermiştir. Sağlam yöntemler ise bazı düşük ve orta kirlenme koşullarında, özellikle temel sürecin korunması amaçlandığında daha iyi sonuç vermiştir. Ek tekrarın yararı; tahminci seçimleri, değerlendirme hedefleri, alternatif süreç geometrileri, hafif yanıt yüzeyi yanlış belirlemesi, küçük örneklem kalibrasyonu ve pozitiflik yaklaşımları altında korunmuştur. Baskı süreci örneği ve yarı-sentetik genişletme, temiz ve hafif bozulma koşullarında aynı nitel örüntüyü göstermiştir. Sabit 27 noktalı tasarım içinde en tutarlı etki tekrar sayısından kaynaklanmış; tercih edilen tahminci ise hedefe ve kirlenme yapısına bağlı kalmıştır. Bulgular, ikili yanıt yüzeyi çalışmalarında tekrar sayısının planlanması ve tahminci seçimi için uygulamaya dönük bir temel sunmaktadır.

Anahtar Kelimeler

• Sağlam parametre tasarımı
• ikili yanıt yüzeyi metodolojisi
• tekrar sayısı
• ölçek tahmini
• kirlenmiş veri

The Effect of Replication Size and Estimator Choice on Dual Response Surface Optimization under Skewed and Contaminated Data

Öz

Within-design-point replication and estimator choice are usually examined separately in dual response surface optimization. This study evaluated their joint influence under clean, skewed, and additively contaminated data using 39 scenarios, 200 Monte Carlo replications, and nested samples at r ∈ {3,5,8,10}. The primary analysis fixed a skewness-adjusted location estimator and compared six classical and robust scale estimators. Decision quality was assessed against full-process and baseline-process targets. Increasing r from 3 to 8 reduced mean full-process regret by 66.9% under clean data, 40.5% under skew-only data, and 38.7%, 24.6%, and 22.5% under low, moderate, and high contamination probabilities. In the joint location–scale analysis, the same transition improved full-process regret in 79 of 80 combinations and baseline-process regret in all 80. No estimator was uniformly best. Standard deviation performed best under clean data and often when contamination was treated as part of the operating process. Robust scale estimators performed better in some low- and moderate-contamination settings, especially when preserving the baseline process was the aim. The benefit of additional replication remained consistent across estimator choices, evaluation targets, alternative process geometries, mild response-surface misspecification, finite-sample calibration, and positivity treatments. A printing-process benchmark and a semi-synthetic extension showed the same qualitative pattern under clean and mild-disturbance conditions. Within the fixed 27-point design, replication had the most consistent effect, while the preferred estimator depended on the target and contamination structure. Together, the results provide practical guidance for planning replication and selecting estimators in dual response surface studies.

Anahtar Kelimeler

• Robust parameter design
• dual response surface methodology
• replication size
• scale estimation
• contaminated data

Kaynakça

1. Genichi Taguchi. Introduction to Quality Engineering: Designing Quality into Products and Processes. Asian Productivity Organization, Tokyo, 1986. ISBN 9789283310839.
2. George E. P. Box. Discussion of “off-line quality control, parameter design, and the Taguchi method”. Journal of Quality Technology, 17(4):189–190, 1985. doi: 10.1080/00224065.1985.11978965.
3. George E. P. Box and K. B. Wilson. On the experimental attainment of optimum conditions. Journal of the Royal Statistical Society: Series B (Methodological), 13(1):1–38, 1951. doi: 10.1111/j.2517-6161.1951.tb00067.x.
4. Raymond H. Myers, Douglas C. Montgomery, and Christine M. Anderson-Cook. Response Surface Methodology: Process and Product Optimization Using Designed Experiments. John Wiley & Sons, Hoboken, NJ, 4th edition, 2016. ISBN 9781118916018.
5. G. Geoffrey Vining and Raymond H. Myers. Combining Taguchi and response surface philosophies: A dual response approach. Journal of Quality Technology, 22(1):38–45, 1990. doi: 10.1080/00224065.1990.11979204.
6. Enrique Del Castillo and Douglas C. Montgomery. A nonlinear programming solution to the dual response problem. Journal of Quality Technology, 25(3):199–204, 1993. doi: 10.1080/00224065.1993.11979454.
7. Dennis K. J. Lin and Wanzhu Tu. Dual response surface optimization. Journal of Quality Technology, 27(1):34–39, 1995. doi: 10.1080/00224065.1995.11979556.
8. Karen A. F. Copeland and Peter R. Nelson. Dual response optimization via direct function minimization. Journal of Quality Technology, 28(3):331–336, 1996. doi: 10.1080/00224065.1996.11979683.

9. Enrique Del Castillo, Shu-Kai Fan, and John Semple. Optimization of dual response systems: A comprehensive procedure for degenerate and nondegenerate problems. European Journal of Operational Research, 112(1):174–186, 1999. doi: 10.1016/S0377-2217(97)00382-2.
10. Rui Ding, Dennis K. J. Lin, and Duan Wei. Dual-response surface optimization: A weighted MSE approach. Quality Engineering, 16(3):377–385, 2004. doi: 10.1081/QEN-120027940.
11. Stephanie M. Pickle, Timothy J. Robinson, Jeffrey B. Birch, and Christine M. Anderson-Cook. A semi-parametric approach to robust parameter design. Journal of Statistical Planning and Inference, 138(1):114–131, 2008. doi: 10.1016/j.jspi.2007.05.018.
12. Sarah E. Burke, Douglas C. Montgomery, Christine M. Anderson-Cook, and Rachel T. Silvestrini. Optimal design of experiments for dual-response systems with normal, binary, and Poisson distributions. Military Operations Research, 28(2):109–134, 2023. URL: https://www.jstor.org/stable/27227017.
13. Matthias H. Y. Tan and Szu Hui Ng. Estimation of the mean and variance response surfaces when the means and variances of the noise variables are unknown. IIE Transactions, 41(11):942–956, 2009. doi: 10.1080/07408170902735418.
14. Chanseok Park and Byung Rae Cho. Development of robust design under contaminated and non-normal data. Quality Engineering, 15(3):463–469, 2003. doi: 10.1081/QEN-120018045.
15. Seong Beom Lee, Chanseok Park, and Byung Rae Cho. Development of a highly efficient and resistant robust design. International Journal of Production Research, 45(1):157–167, 2007. doi: 10.1080/00207540600649202.
16. Gregory L. Boylan and Byung Rae Cho. Comparative studies on the high-variability embedded robust parameter design from the perspective of estimators. Computers & Industrial Engineering, 64(1):442–452, 2013. doi: 10.1016/j.cie.2012.10.012.
17. Peter J. Huber. Robust estimation of a location parameter. The Annals of Mathematical Statistics, 35(1):73–101, 1964. doi: 10.1214/aoms/1177703732.
18. Frank R. Hampel, Elvezio M. Ronchetti, Peter J. Rousseeuw, and Werner A. Stahel. Robust Statistics: The Approach Based on Influence Functions. John Wiley & Sons, New York, 1986. ISBN 9780471829218. doi: 10.1002/9781118186435.
19. David A. Lax. Robust estimators of scale: Finite-sample performance in long-tailed symmetric distributions. Journal of the American Statistical Association, 80(391):736–741, 1985. doi: 10.1080/01621459.1985.10478177.
20. Peter J. Rousseeuw and Christophe Croux. Alternatives to the median absolute deviation. Journal of the American Statistical Association, 88(424):1273–1283, 1993. doi: 10.1080/01621459.1993.10476408.
21. Peter J. Rousseeuw and Sabine Verboven. Robust estimation in very small samples. Computational Statistics & Data Analysis, 40(4):741–758, 2002. doi: 10.1016/S0167-9473(02)00078-6.
22. John A. Randal. A reinvestigation of robust scale estimation in finite samples. Computational Statistics & Data Analysis, 52(11):5014–5021, 2008. doi: 10.1016/j.csda.2008.04.016.
23. Chanseok Park, Haewon Kim, and Min Wang. Investigation of finite-sample properties of robust location and scale estimators. Communications in Statistics—Simulation and Computation, 51(5):2619–2645, 2022. doi: 10.1080/03610918.2019.1699114.
24. Gregory L. Boylan and Byung Rae Cho. Studies on the effects of estimator selection in robust parameter design under asymmetric conditions. Quality and Reliability Engineering International, 29(4):571–582, 2013. doi: 10.1002/qre.1406.
25. Elif Kozan. A new robust estimation approach to dual response surface models. Duzce University Journal of Science and Technology, 13(3):1089–1097, 2025. doi: 10.29130/dubited.1537921.
26. Agah Kozan. An interval-based robust response surface design using the W24 estimator for asymmetric and contaminated data. Symmetry, 18(4):644, 2026. doi: 10.3390/sym18040644.
27. Norman J. Johnson. Modified t tests and confidence intervals for asymmetrical populations. Journal of the American Statistical Association, 73(363):536–544, 1978. doi: 10.1080/01621459.1978.10480051.
28. Melis Zeybek. Confidence interval based quality improvement for non-normal responses. Celal Bayar University Journal of Science, 15(2):199–204, 2019. doi: 10.18466/cbayarfbe.518736.
29. Melis Zeybek. Interval robust design under contaminated and non-normal data. Communications in Statistics—Theory and Methods, 49(22):5406–5418, 2020. doi: 10.1080/03610926.2019.1710198.
30. Linhan Ouyang, Yizhong Ma, Jai-Hyun Byun, Jianjun Wang, and Yiliu Tu. An interval approach to robust design with parameter uncertainty. International Journal of Production Research, 54(11):3201–3215, 2016. doi: 10.1080/00207543.2015.1078920.
31. Steven G. Gilmour and Luzia A. Trinca. Optimum design of experiments for statistical inference. Journal of the Royal Statistical Society: Series C (Applied Statistics), 61(3):345–401, 2012. doi: 10.1111/j.1467-9876.2011.01000.x.
32. Gregory L. Boylan, Paul L. Goethals, and Byung Rae Cho. Robust parameter design in resource-constrained environments: An investigation of trade-offs between costs and precision within variable processes. Applied Mathematical Modelling, 37(4):2394–2416, 2013. doi: 10.1016/j.apm.2012.05.017.
33. Stephen J. Walsh, Thomsen B. Bolton, and John J. Borkowski. Generating optimal designs with user-specified pure replication structure. Quality and Reliability Engineering International, 40(4):2019–2047, 2024. doi: 10.1002/qre.3506.
34. Alan M. Gross. Confidence interval robustness with long-tailed symmetric distributions. Journal of the American Statistical Association, 71(354):409–416, 1976. doi: 10.1080/01621459.1976.10480359.
35. David C. Hoaglin, Frederick Mosteller, and John W. Tukey, editors. Understanding Robust and Exploratory Data Analysis. John Wiley & Sons, New York, 1983. ISBN 9780471097778.
36. Timothy C. Beers, Kevin Flynn, and Karl Gebhardt. Measures of location and scale for velocities in clusters of galaxies—a robust approach. The Astronomical Journal, 100:32–46, 1990. doi: 10.1086/115487.
37. Alan M. Gross. Confidence intervals for bisquare regression estimates. Journal of the American Statistical Association, 72(358):341–354, 1977. doi: 10.1080/01621459.1977.10481001.
38. Frank Wilcoxon. Individual comparisons by ranking methods. Biometrics Bulletin, 1(6):80–83, 1945. doi: 10.2307/3001968.
39. Sture Holm. A simple sequentially rejective multiple test procedure. Scandinavian Journal of Statistics, 6(2):65–70, 1979. URL: https://www.jstor.org/stable/4615733.
40. Bradley Efron and Robert J. Tibshirani. An Introduction to the Bootstrap. Chapman & Hall, New York, 1993. ISBN 9780412042317. doi: 10.1201/9780429246593.
41. Tim P. Morris, Ian R. White, and Michael J. Crowther. Using simulation studies to evaluate statistical methods. Statistics in Medicine, 38(11):2074–2102, 2019. doi: 10.1002/sim.8086.