One-step-ahead predictive quantile residuals (Masarotto and Varin, 2012,
2017) from a model fitted by fit_copula(): the standardized innovations
evaluated at the maximum-likelihood estimates. Under correct specification
and known parameters, these are i.i.d. standard normal; see the paper's
Proposition 1 for the finite-sample statement.
Arguments
- fit
A list as returned by
fit_copula().
Value
A list with components r (the residuals), mu, shape (the
fitted marginal parameters) and eps (the untransformed latent scale).