0
$\begingroup$

Continuous marginal distributions have a potentially unbounded support. (does unbounded support mean $[0,\infty)$ as much as it could mean $(-\infty,\infty)$?)

Is it true then that, on the other hand, a copula:

  1. bypasses the estimation of the margins, and
  2. the copula has a support in $[0,1]^2$ because its margins are uniform?

Why would a range of $[0,1]^2$ be desirable over $[0,1]$. What is the reason for the exponent $2$ with respect to uniform marginals?

$\endgroup$
3
$\begingroup$

Is it true then that, on the other hand, a copula (1) bypasses the estimation of the margins, and (2) the copula has a support in $[0,1]^2$ because its margins are uniform?

(1) It is not true that employing a copula makes estimation of the margins unnecessary. Regardless of what model you choose for the bivariate distribution (copula-based or something different), if the margins are not explicitly given (assumed to be known), they have to be estimated. And in most real-world problems, I would not expect the margins to be given.

(2) Yes.

Why would a range of $[0,1]^2$ be desirable over $[0,1]$. What is the reason for the exponent $2$ with respect to uniform marginals?

Given a bivariate distribution, there are two random variables we are dealing with. Accordingly, the language of mathematics and probability requires two arguments in the joint distribution function as well as in the copula function. It is not that we can choose between $[0,1]^2$ and $[0,1]$ freely, so the question of desirableness is not really relevant.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.