Skip to main content
. 2021 Jul 14;15:684423. doi: 10.3389/fncom.2021.684423

TABLE 2.

Performance of the different regressors evaluated in the study.

Linear Pchip Spline Nearest




MSPE R2 MSPE R2 MSPE R2 MSPE R2
RR 24.96 ± 3.61 0.76 ± 0.03 24.14 ± 3.35 0.76 ± 0.03 24.56 ± 3.41 0.76 ± 0.03 23.65 ± 3.26 0.77 ± 0.03
RF 10.62 ± 2.12 0.92 ± 0.02 11.11 ± 2.37 0.91 ± 0.02 11.24 ± 2.53 0.91 ± 0.02 11.97 ± 2.91 0.90 ± 0.02
MLP 23.09 ± 3.63 0.77 ± 0.03 22.20 ± 3.37 0.78 ± 0.02 20.14 ± 2.91 0.80 ± 0.02 23.56 ± 3.44 0.73 ± 0.04
SVR 18.97 ± 3.50 0.80 ± 0.04 18.67 ± 3.43 0.80 ± 0.04 17.84 ± 3.66 0.81 ± 0.04 21.46 ± 3.55 0.80 ± 0.03

Columns display the average mean squared percentage error and Pearson correlation coefficient for each of the four interpolation methods applied. For both metrics we have included also the standard error of the mean (SEM). Rows represent the different regressors that we evaluated. The bold values represent the highest performance yielded in terms of MSPE and R2.