デジタルの標準SN比による欠測値対策の評価

抄録

Missing data preclude calculation of the Mahalanobis distance in the Mahalanobis-Taguchi system. The feasibility of evaluation using a digital standard S/N ratio in place of missing data was therefore studied, taking health examination data as an example. The significance of this study is that it showed when there are missing values in the data, the loss of discriminatory power of a health examination using the Mahalanobis distance can be minimized by use of mean values of healthy subjects as replacement data. The results of this study offer suggestions for dealing with missing data when the MT system is applied in a variety of other fields as well as in health examination.