Fig. 5

K-means clustering result with the best classification performance for each number of the variables used. Biplots consisting of the principal components are used to indicate the distribution of clustered samples (a, c, e), and the corresponding quantities of lipid induction amount for each cluster are expressed in boxplots (b, d, e). In the case of just one variable used, the axes represent the corresponding condition and the lipid induction (e). Three conditions (a, b), two conditions (c, d), and one condition used (e, f). CH Chl a