Fig. 11From: Deriving amplification factors from simple site parameters using generalized regression neural networks: implications for relevant site proxiesVariation in root-mean-square error, standard deviation of residuals \(\varepsilon_{\text{GRNN}} \left( {\theta ,\nu_{i} } \right),\) for various GRNN models involving various sets of input parameters (indicated with different colors) compared to the initial variability \(\sigma_{0} \left( {\theta ,\nu_{i} } \right)\) for RP–NF (a, top), NP–NF (b, middle) and TP–NF (c, bottom). Data are displayed as a function of normalized frequency \(\nu = f/f_{0}\) Back to article page