Copula

Overview

Contains many tools useful for copula modeling in Matlab that do not exist directly in the Statistics and Machine Learning toolbox. Highlights are:

• Smooth empirical copula density estimation via Beta-Kernels for any D>=2.
• Empirical copula function estimation for any D>=2
• Sampling from calculated empirical copula for D>=2
• Clayton/Frank/Gumbel copula PDF and sampling for D>=2

Directory Structure

• algorithms/ - contains the core copula algorithms.

| File | Description | | --- | --- | | claytoncopulapdf.m | Computes the Clayton Copula's PDF for D>=2 | | claytoncopularnd.m | Samples from a D>=2 Clayton Copula | | computeEmpiricalDiscreteProb.m | Computes empirical multinomial distribution | | continueRv.m | Continues realizations of a discrete RV (see http://dx.doi.org/10.1016/j.jmva.2004.01.004) | | empcopulaval.m | Computes value of an empirical copula at a specified point in unit hypercube | | empcopulapdf.m | Computes empirical copula density given pseudo-observations | | empcopulacdf.m | Computes empirical copula function given pseudo-observations | | empcopularnd.m | Generates samples from an empirical copula | | estMteDensity.m | KDE with trucanted exponential distribution | | frankcopulapdf.m | Computes the Frank Copula's PDF for D>=2 | | frankcopularnd.m | Samples from a D>=2 Frank Copula | | gumbelcopulapdf.m | Computes the Gumbel Copula's PDF for D>=2 | | gumbelcopularnd.m | Samples from a D>=2 Gumbel Copula | | hyperFunctionError.m | Computes error between two hyper functions of the same dimensionality | | log1mexp.m | Convenience function for log(1-exp(a)) | | logserrnd.m | Samples from the Log-Series distribution | | pobs.m | Computes pseudo-observations for a given (multivariate) random vector | | stable1rnd.m | Samples from the Stable Distribution | | etstablernd.m | Samples from the Exponentially Tilted Distribution | | logrnd.m | Samples from the Log Distribution |

• simulations/ - contains simulation code which uses the algorithms developed