The support of the GEV distribution changes dependent on the value of the shape parameter: \xi > 0: \hspace{1cm} y \in \left[\mu - \frac{\sigma}{\xi}, +\infty\right) \hspace{0.25cm} \implies \hspace{0.25cm} \text{min(y)} \geq \mu - \frac{\sigma}{\xi} \xi = 0: \hspace{1cm} y \in (-\infty, +\infty) …

We ended up coming back to this problem nearly a year later and working out a solution, so thought I’d post here just in case it’s relevant to someone. The support of the GEV depends on all three parameters and the data: \begin{equation} \{ y : 1 + \xi(y-\mu)/\sigma > 0 \} \end{equation} where -…

Aberrant behavior in dynamically bounded parameters, reparameterized GEV distribution

Modeling

martinmodrak June 25, 2021, 9:26pm 4

I just saw this on Twitter and it looks highly relevant: [2106.13110] Practical strategies for GEV-based regression models for extremes

2 Likes

Too more divergent transitions and too large R-hat in bayesian hierarchical generalized extreme value model

Topic		Replies	Views
Variable bounds depending on parameter Modeling fitting-issues , specification	8	2456	November 2, 2017
Log(0) error, when defining shape parameter with normal distribution containing regression parameters in non-stationary GEV Modeling rstan , reparametrization	5	485	February 14, 2023
Implementing Generalized Exteme Value distribution (GEV) constraints Modeling	0	449	October 8, 2022
Reparameterizing GEV Distribution Modeling fitting-issues	6	2691	December 7, 2017
Fitting a basic uniform distribution -- what am I missing? Modeling fitting-issues	11	871	December 7, 2020

Aberrant behavior in dynamically bounded parameters, reparameterized GEV distribution

Related topics