Copula: Log-likelihood, AIC, BIC

Question

Elena 2013-10-22

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/91098-copula-log-likelihood-aic-bic

编辑： kantha vel 2018-9-10

Dear All, I am trying to estimate the best copula and have read that one approach is to compare log-likelihood and some information criteria (AIC, BIC). But I can not calculate them, also in the examples provided there is no info how to proceed. Help please... Would be very grateful! Thanks in advance.

1 个评论
显示 -1更早的评论隐藏 -1更早的评论

Antonio 2014-9-20

I have the same question, I have two dataset that are distributed following a Normal Distribution, after using the normcdf to make them Uniform, afterwards, I use the copulafit comamand to fit with the different families of copulas, and now I would like to check which is the one that adapt better, in order to measure the dependence among the data.

I guess that AIC and BIC are the best way to deal with this, but I do not know how to implement them.

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

M Bateni 2016-6-23

1
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/91098-copula-log-likelihood-aic-bic#answer_226511

You can go to the body of built-in file 'copulafit' by typing 'edit copulafit' in the command window. At the end of the body of that function, there are some sub-functions starting with "negloglike" like 'negloglike_clayton'. By using those functions out of 'copulafit', you can have negative likelihood values for different copula families. having this value, you can easily calculate AIC or BIC (maybe using 'aicbic' function).