What steps will reproduce the problem?
1. Fit GPD with phat = fitgenpar(data,options)
2. Calculation of CI by [x,xlo,xup] = invgenpar(F,phat,options) for
What is the expected output? What do you see instead?
I want to calculate return levels and the corresponding CI bounds.
First, I do a GPD fit with phat = fitgenpar(data,options).
that gives me:
1. the array with gpd-distribution phat.params
2. the upper and lower CI phat.upperbound and phat.lowerbound
Now, I calculate the return level for Tr (first approach):
xr = invgenpar(1/(lambda*Tr), phat.params(1), phat.params(2), phat.params(3),
'lowertail',false);
The CI bounds are:
xru = invgenpar(1/(lambda*Tr), phat.upperbound(1), phat.upperbound(2),
phat.upperbound(3), 'lowertail',false);
xru = invgenpar(1/(lambda*Tr), phat.lowerbound(1), phat.lowerbound(2),
phat.lowerbound(3), 'lowertail',false);
The same could be done (if I understand right) with (second approach):
[xr,xrl,xru] = invgenpar(1/(lambda*Tr, phat, 'lowertail',false)
Now, as expected, the values for xr are the same in the first and in the second
approach. But, the values xru and xrl (CI bounds) differ. The second appoach
produces very wide bands. How comes?
What version of the product are you using? On what operating system?
Newest WAFO 2.5 on MATLAB R2012a on Win7.
Please provide any additional information below.
Original issue reported on code.google.com by mkska...@gmail.com on 24 Apr 2013 at 12:12
Original issue reported on code.google.com by
mkska...@gmail.comon 24 Apr 2013 at 12:12