Compute composite loss value for epsilon-insensitive type function
loss2_ccsvm(y, f, weights, cfun, dfun, s, eps, delta=0.0001)response variable values
fitted values of y
observation weights, same length as y
integer from 1-8, concave function as in ccsvm_fit
integer value, only dfun=2 is implemented for now. Convex function as in ccsvm_fit
tuning parameter of cfun. s > 0 and can be equal to 0 for cfun="tcave".
a small positive number provided by user only if cfun="gcave" and 0 < s <1
non-negative parameter for epsilon-insensitive loss
Weighted loss values
For large s values, the loss can be 0 with cfun=2,3,4, or "acave", "bcave", "ccave".
Zhu Wang (2020) Unified Robust Estimation via the COCO, arXiv e-prints, https://arxiv.org/abs/2010.02848