Hi, I am trying to fit a Gumbel EV distribution to data of the Maximums of values generated from a normal distribution. Plotted a histogram and would like to fit the Gumbel and see how close the fit is, however the problem is that I do not know the parameters a - location and b - scale respectively. Can anyone help me with these (formulas for paremeters) without having to maximize the derivative of the log of the likelihood? Same thing with the values of the parameters of the Frechet as I would like to fit a Frechet to an underlying student t - distribution and compare. I.e in a similar way as u_n and sigma have been estimated for the Gumbel with an underlying exponential in Q&A Bank 4.17. A little help will go a long way.
Stackexchange gives a formula for the MLE: http://stats.stackexchange.com/ques...imators-for-parameters-in-gumbel-distribution However this paper cautions against MLE performance with small samples, so use at your own risk: http://revistas.ucr.ac.cr/index.php/matematica/article/viewFile/259/239
