Is it possible to estimate the shape, scale, and location parameters of a generalized extreme distribution (gev) if I just know the mean, variance, and median of a given data set (i.e., no raw data available - - just its descriptive statistics)?
If I undertood well your question, basicaly, you you have to solve a system of equations with three unknowns;
See the attached picture . so, you can replace the estimated values of : Mean, Median and variance to calculate those of location, scale and shape paraemters.
The formulas of the mean and the variance of generalized extreme value distribution are already calculated, I think you can use them to found the value of the parameters of generalized extreme value distribution. You can found all these formulas in my thesis.Thesis Théorie des Valeurs Extrêmes : Application au Calcul de Risques
See the link:https://www.google.com/search?client=firefox-b-1-d&ei=OJaLX8nUEtTEtQaj-KFQ&q=How+to+find+the+moments+of+a+generalized+extreme+value+distribution&oq=How+to+find+the+moments+of+a+generalized+extrem+value+distribution&gs_lcp=CgZwc3ktYWIQDFAAWABgrLkDaABwAHgAgAEAiAEAkgEAmAEAqgEHZ3dzLXdpeg&sclient=psy-ab&ved=0ahUKEwjJ0LLn-rzsAhVUYs0KHSN8CAoQ4dUDCAw
You can use the maximum likelihood estimator, or the probabilistic weighted moments, or other methods. The choice depend on a lot of situations inherent to the data. But, first try MLE.