Firstly, each dataset is analyzed with regression analysis of a statistical method. Secondly, discriminant analysis (DA) sets a discriminant criterion from the results by the regression analysis, and classifies the group by the criterion. Linear discriminant formula calculates a value which can classify the group. Then, each DA value is divided into the positive group or the negative group. In the regression analysis, single regression analysis in case of one explanatory variable is available, and partial least squares (PLS) regression analysis in case of more than one explanatory variable is available because a problem of multicollinearity may be occurred and it is difficult to decide the group border when one explanatory variable has correlation with another explanatory variable. When DA using multiple regression analysis is executed in a case of Dataset No. 4 shown in Table 2, Fig. 2 shows that it is difficult to decide the group boundary because there is variability around the threshold which value is “0”. Instead of multiple regression analysis, PLS regression analysis is introduced. As a result, Fig. 3 shows that it is easy to decide the group boundary at 34.7 %.