A copula is a multivariate probability distribution for which the marginal probability distribution of each variable is uniform. It is a very popular form of dependency modelling between many disparate processes with different underlying distributions.
Imagine you are in a factory which makes a Nintendo Switch. To manufacture a Switch, you need parts coming from 3 different processes (they could even be manufactured in different countries). Let’s call them \(A\), \(B\), \(C\).
If the failure rate of each process repectively are (for example) given by an Exponential, a Weibull and a Beta distribution individually, how do we model the chance of them failing together? (That would be catastrophe for all Pokémons!)
If only we could assume that all processes were independent of one another. Or if only we could assume the failure rates of all processes were given by a Normal distribution, then we could dump a covariance matrix that explains the processes’ relationships.
But hope is not lost, (Pokémons rejoice!) this is where copulas (and Copulae) comes in.
Presently, only elliptical and Archimedean copulas have been implemented. I’m working on putting more common copulas into the package. I’ll also be adding some common charts and statistical goodness-of-fit functions that can help the lay person more quickly learn the concepts. I struggled a bunch trying to pick this up, hope you’ll have an easier time!