Predictions by different models at medium noise level - Cross validation. The blue stars denote the data points whereas the magenta squares show the SPCA predictions when the corresponding data points were excluded as test set elements. The red and the green lines show the reconstructed time profiles of the metabolites by using the true and the simplified models for different test sets, respectively. The magnifying window in the lower right hand side of the figure shows the differences of the reconstructed time profiles for different test sets. There, the deviation between the lines of the same color (obtained by using different test sets but the same model description each time) can be seen in great detail.