I am using copulas as a tool to better understand the nuanced dependency amongst education outcomes, e.g. dropout and approved.
I have been able to use bicop command to run all the estimations and define the best model.
Code:
local maxll=minfloat() * * Mixture <- specifies the marginal distribution of each residual * none specifies each marginal distribution as an N(0, 1) form; foreach cop in gaussian frank clayton gumbel joe indep { local xvars boy nowi urb age govaid pc element sroom lib sci sports tage tagesd stu_staff_ratio bicop status edrop `xvars', copula(`cop') mixture(none) estimates store `cop' if e(ll)> `max11' e(converged) { local `max11'=e(11) local bestcop="`cop'" matrix bestb=e(b) } } * Estimation summary statistics estimates stats _all
Well, how can I get the Kendall tau measure of correlation?
Also, is there a way for me to get the residuals from both equations? I tried
Code:
predict varres, r equation (#1)
Thank you all.
Max
0 Response to Empirical Joe Copula -
Post a Comment