Total Quality Science
Online ISSN : 2189-3195
ISSN-L : 2189-3195
Current issue
Displaying 1-3 of 3 articles from this issue
  • Yukiya Tomizawa, Hiroki Iwamoto, Yasushi Nagata
    Article type: research-article
    2024 Volume 10 Issue 2 Pages 33-43
    Published: December 24, 2024
    Released on J-STAGE: December 24, 2024
    JOURNAL OPEN ACCESS

    Robust parameter design is a central technique of the Taguchi method, wherein parameters are designed such that output fluctuations are reduced even when the environmental and operating conditions change marginally. Generally, it is necessary to level the noise factor however this is often difficult to assign. It is sometimes possible to passively measure the value of the noise factor that is, when its value of the noise factor can be observed as a covariate.

    Koyano (2013) proposed a robust parameter design method that uses propensity scores when multiple noise factors are observed as covariates. Goto et al. (2015) proposed a robust parameter design method when multiple noise factors are observed as covariates, thus extending the method of Hirano and Miyakawa (2007).

    The aforementioned studies only examined cases wherein each control factor had two levels. In this study, we proposed two extensions of the above methods for multilevel control factors: extended Method 1, based on the Koyano (2013) method, and Method 2, based on the Goto et al. (2015) method. Simulations were performed based on the aforementioned studies to evaluate the performance of each method. Method 1 was not valid under any of the simulation conditions. Conversely, Method 2 showed validity regardless of the conditions.

    Download PDF (306K)
  • Riku Fujii, Hiroki Iwamoto, Yasushi Nagata
    Article type: research-article
    2024 Volume 10 Issue 2 Pages 44-53
    Published: December 24, 2024
    Released on J-STAGE: December 24, 2024
    JOURNAL OPEN ACCESS

    Taguchi’s T-method is one of the Mahalanobis-Taguchi (MT) system proposed by Gen-ichi Taguchi. It is used for various forecasting purposes. The T-method is often compared to multiple regression analysis because the data type applied is the same in both methods. One advantage of the T-method over multiple regression analysis is its ability to deal with missing data. Unlike multiple regression analysis, the T-method can be analyzed without removing missing samples. In this study, we propose an improved method that further improves the accuracy of missing data while maintaining the features of the T-method. Specifically, we first improve the T-, Ta-, and Tb-methods using bootstrapping for data with missing values and then investigate a new method for dealing with missing data. Next, by calculating the prediction accuracy using simulations under various models, we examine whether there is a trend toward superiority or inferiority in multiple regression analysis, the T-, Ta-, and Tb-methods as well as in the improved methods over the T-, Ta-, and Tb-methods in the case of missing data. The results show that the improved Tb-method is more accurate than the conventional one, regardless of missing mechanisms. The improved T- and Ta-methods are more accurate in some cases.

    Download PDF (418K)
  • Takuya Matsumoto, Hideki Nagatsuka
    Article type: research-article
    2024 Volume 10 Issue 2 Pages 54-63
    Published: December 24, 2024
    Released on J-STAGE: December 24, 2024
    JOURNAL OPEN ACCESS

    The stable distribution is used in various fields because it is a generalization of the normal, Cauchy, Lévy distributions and it has desirable properties such as infinitely divisible distribution. However, conducting the maximum likelihood methods for the stable distributions is very difficult since the probability density function has the complicated form and it is difficult to calculate it. Meanwhile, the characteristic function (cf) of the stable distribution has closed-form expressions. So, the methods of parameter estimation for the stable distribution, based on the cf, which is referred to as the minimum distance estimation (MDE) method has been proposed. However, in the MDE method, we found that the smaller the value of the parameter α, the greatly larger the bias and RMSE of the estimators in this method.

    Motivated by the above-mentioned problem, in this article, we propose the new parameter estimation method based on the MDE method, for the stable distribution, that is robust to outliers. To remedy this problem, The key idea of the proposed method is to use a robust minimum distance estimation. Through Monte Carlo simulations, we evaluate the performance of the proposed estimators compared with the existing MDE method in terms of bias and RMSE and show the proposed method performs satisfactorily for all α, even the case of the small α. Furthermore, in an illustrative example, we demonstrate the proposed method outperforms the existing method.

    Download PDF (525K)
feedback
Top