An Application of Generalized Inverse Regression Estimator to Taguchi's T Method  

Oleh:

Kawada, Hiroshi ; Nagata, Yasushi

Jenis: Article from Proceeding
Dalam koleksi: 12th ANQ Congress in Singapore, 5-8 Agustus 2014, page 1-8.
Topik: MT system; T method; Prediction; Generalized inverse regression estimator; Monte Carlo simulation
Fulltext: QP1-1.2-P0244.pdf (719.26KB)

Isi artikel

Taguchi’s T method is one of the MT (Mahalanobis-Taguchi) system. T method is technique for prediction, and the application in various fields such as economic forecast or the demand prediction is possible. The idea of T method is to prepare many weak learners. And then, through weighted integration of predicted values obtained for each weak learner by signal to noise (S/N) ratio, an overall estimated output value is obtained. Now, if the accuracies of the estimated value by each item are raised, it will be thought that the accuracy of the combined estimated value finally outputted can be raised. Therefore, in this report, we intend to improve the accuracies of the estimated values derived by each item. In this process, we propose a method that introduced the concept of generalized inverse regression estimator in each item. By Monte Carlo simulation, we compare the prediction accuracy of the proposed method and that of the existing method under some conditions. And, it turned out that prediction accuracy of the proposal method is better in many cases. In particular, when the number of samples is small, a large improvement in prediction accuracy is observed. From the above, we think that the proposal by this report is effective

Opini Anda

Klik untuk menuliskan opini Anda tentang koleksi ini!

Kembali