As seen from Figure 3A, we obtained a best fitting value of R=2.66 nm (red curve).
This approach with one neuron demand in hidden layer exhibits the fitting value of R2=0.96, which is better than linear regression model.
As expected, it was found that the dynamic correction factor can make the fitting value of prediction model close to the real value infinitely, and almost achieved linear fitting perfectly.
Briefly, the profile of each metabolite under each condition was fitted to a polynomial model of maximum degree 3. The curves of each genotype were statistically compared taking into account the fitting value and correcting the p-value (Benjamini Hochberg method).
The fitting value is normalized by the fitting value of the first iteration of each period band.
The modelling error refers to the difference between the model fitting value and the true value.
The most fitting value of ζ is 0.2, while the best value of κ is 0.4.
The fitting value of σdc/σop of around 300 was attributed to the spin-coating process.
The fitting value of σdc/σop of 299.3 was attributed to the spin-coating process.
The fitting value of straight slope is shown in Table 1.
The suggested value of κ (equal to 0.4) is best fitting value for the stiffness trend at increasing load.
