copula_transformation.RdTransforms the entries an arbitrary copula so they can be compatible with the first two moments of a timeseries.
copula_transformation(x, copula)
| x | A multivariate (non-tidy) dataset. |
|---|---|
| copula | A multivariate copula. |
A tabular (non-tidy) tibble.
Under the hood, the copula transformation is a 4-step recipe:
1. Normalize the copula to have a standard normal distribution;
2. Operate an affine transformation on the normalized variables;
3. Do the CMA Separation step on the new rotated variables;
4. Extract the copula component that results from step 3.
x <- matrix(diff(log(EuStockMarkets)), ncol = 4) sep <- cma_separation(x) clayton <- fit_copula_clayton(sep) gen <- generate_copulas(clayton, 10) copula_transformation(x, gen) #> # A tibble: 10 × 4 #> ...1 ...2 ...3 ...4 #> <dbl> <dbl> <dbl> <dbl> #> 1 0.909 0.909 0.545 0.545 #> 2 0.364 0.818 0.727 0.818 #> 3 0.182 0.727 0.182 0.364 #> 4 0.636 0.273 0.818 0.909 #> 5 0.0909 0.0909 0.0909 0.0909 #> 6 0.545 0.455 0.455 0.455 #> 7 0.818 0.636 0.909 0.636 #> 8 0.455 0.364 0.636 0.182 #> 9 0.273 0.182 0.364 0.273 #> 10 0.727 0.545 0.273 0.727