Generalized extreme value distribution
From Wikipedia, the free encyclopedia
| Probability density function | |
| Cumulative distribution function | |
| Parameters | <math>\mu \in [-\infty,\infty] \,</math> location (real) <math>\sigma \in (0,\infty] \,</math> scale (real) |
|---|---|
| Support | <math>x>\mu-\sigma/\xi\,\;(\xi > 0)</math> <math>x<\mu-\sigma/\xi\,\;(\xi < 0)</math> |
| Probability density function (pdf) | <math>\frac{1}{\sigma}(1\!+\!\xi z)^{-1/\xi-1}e^{-(1\!+\!\xi z)^{-1/\xi |
| Cumulative distribution function (cdf) | {{{cdf}}} |
| Mean | {{{mean}}} |
| Median | {{{median}}} |
| Mode | {{{mode}}} |
| Variance | {{{variance}}} |
| Skewness | {{{skewness}}} |
| Excess Kurtosis | {{{kurtosis}}} |
| Entropy | {{{entropy}}} |
| mgf | {{{mgf}}} |
| Char. func. | {{{char}}} |
where <math>z=\frac{x-\mu}{\sigma}</math>|
cdf =<math>e^{-(1+\xi z)^{-1/\xi}}</math>|
mean =|
median =<math>\frac{\ln^{-\xi}(2)-1}{\xi}</math>|
mode =<math>\frac{(1+\xi)^{-\xi}-1}{\xi}</math>|
variance =|
skewness =|
kurtosis =|
entropy =|
mgf =|
char =|
}}
In probability theory and statistics, the generalized extreme value distribution (GEV) is a family of continuous probability distributions developed within extreme value theory to combine the Gumbel, Fréchet and Weibull families also known as type I, II and III extreme value distributions. Its importance arises from the fact that it is the limit distribution of the maxima of a sequence of independent and identically distributed random variables. Because of this, the GEV is used as an approximation to model the maxima of long (finite) sequences of random variables.
Contents |
[edit] Specification
The generalized extreme value distribution has cumulative distribution function
- <math>F(x;\mu,\sigma,\xi) = \exp\left\{-\left[1+\xi\left(\frac{x-\mu}{\sigma}\right)\right]^{-1/\xi}\right\}</math>
for <math>1+\xi(x-\mu)/\sigma>0</math>, where <math>\mu\in\mathbb R</math> is the location parameter, <math>\sigma>0</math> the scale parameter and <math>\xi\in\mathbb R</math> the shape parameter.
The density function is, consequently
- <math>f(x;\mu,\sigma,\xi) = \frac{1}{\sigma}\left[1+\xi\left(\frac{x-\mu}{\sigma}\right)\right]^{-1/\xi-1} </math>
- <math>\exp\left\{-\left[1+\xi\left(\frac{x-\mu}{\sigma}\right)\right]^{-1/\xi}\right\}</math>
again, for <math>1+\xi(x-\mu)/\sigma>0</math>.
[edit] Link to Fréchet, Weibull and Gumbel families
The shape parameter <math>\xi</math> governs the tail behaviour of the distribution, the sub-families defined by <math>\xi\to 0</math>, <math>\xi>0</math> and <math>\xi<0</math> correspond, respectively, to the Gumbel, Fréchet and Weibull families, whose cumulative distribution functions are reminded below.
- Gumbel or type I extreme value distribution
- <math> F(x;\mu,\sigma)=e^{-e^{-(x-\mu)/\sigma}}\;\;\; for\;\; x\in\mathbb R </math>
- Fréchet or type II extreme value distribution
- <math> F(x;\mu,\sigma,\alpha)=\begin{cases} 0 & x\leq \mu \\ e^{-((x-\mu)/\sigma)^{-\alpha}} & x>\mu \end{cases}</math>
- Weibull or type III extreme value distribution
- <math> F(x;\mu,\sigma,\alpha)=\begin{cases} e^{-(-(x-\mu)/\sigma)^{\alpha}} & x<\mu \\ 1 & x\geq \mu \end{cases}</math>
where <math>\sigma>0</math> and <math>\alpha>0</math>.
Remark I: For reliability issues the Weibull distribution is used with the variable <math> t = \mu - x </math>, the time, which is strictly positive. Thus the support is positive - in contrast to the use in extreme value theory.
Remark II: Be aware of an important distinctive feature of the three extreme value distributions: The support is either unlimited, or it has an upper or lower limit.
One can link the type I to types II and III the following way: if the cumulative distribution function of some random variable <math>X</math> is of type II: <math>F(x;0,\sigma,\alpha)</math>, then the cumulative distribution function of <math>\ln X</math> is of type I, namely <math>F(x;\ln \sigma, 1/\alpha)</math>. Similarly, if the cumulative distribution function of <math>X</math> is of type III: <math>F(x;0,\sigma,\alpha)</math>, the cumulative distribution function of <math>\ln X</math> is of type I: <math>F(x;\ln \sigma, -1/\alpha)</math>.
[edit] Extremal types theorem
Credit for the extremal types theorem (or convergence to types theorem) is given to Gnedenko (1948), previous versions were stated by Fisher and Tippett in 1928 and Fréchet in 1927.
Let <math>X_1,X_2\ldots</math> be a sequence of independent and identically distributed random variables, let <math>M_n=\max\{X_1,\ldots,X_n\}</math>. If two sequences of real numbers <math>a_n, b_n</math> exist such that <math>a_n>0</math> and
- <math> \lim_{n \to \infty}P\left(\frac{M_n-b_n}{a_n}\leq x\right) = F(x)</math>
then if <math>F</math> is a non degenerate distribution function, it belongs to either the Gumbel, the Fréchet or the Weibull family.
Clearly, the theorem can be reformulated saying that <math>F</math> is a member of the GEV family.
It is worth noting that the result, which is stated for maxima, can be applied to minima by taking the sequence <math>-X_n</math> instead of the sequence <math>X_n</math>.
For the practical application this theorem means: For samples taken from a well behaving, arbitrary distribution <math>X</math> the resulting extreme value distribution <math>M_n</math> can be approximated and parametrised with the extreme value distribution with the appropriate support.
Thus the role of extremal types theorem for maxima is similar to that of central limit theorem for averages. The latter states that the limit distribution of arithmetic mean of a sequence <math>X_n</math> of random variable is the normal distribution no matter what the distribution of the <math>X_n</math>, The extremal types theorem is similar in scope where maxima is substituted for average and GEV distribution is substituted for normal distribution.
[edit] References
- Leadbetter, M.R., Lindgreen, G. and Rootzén, H. (1983). Extremes and related properties of random sequences and processes. Springer-Verlag. ISBN 0-387-90731-9.
- Resnick, S.I. (1987). Extreme values, regular variation and point processes. Springer-Verlag. ISBN 0-387-96481-9.
- Coles, Stuart (2001). An Introduction to Statistical Modeling of Extreme Values,. Springer-Verlag. ISBN 1-85233-459-2.
