Extreme value analysis in R with extRemes and in2extRemes

extRemes (version >= 2.0) and in2extRemes1 are R packages for statistically analyzing extreme values of a data set. in2extRemes is a point-and-click software tool that operates many of the command-line functions from the extRemes (version >= 2.0) package.2

Sign up to receive updates when new package versions are submitted to CRAN (note that this list does not allow members or others to send emails to it, only the list administrators).

You may want to join the User Group for extRemes. The user group is intended for users to discuss issues about EVA and/or extRemes/in2extRemes with each other. Issues regarding bugs, etc., should still be sent to the maintainer (currently, Eric Gilleland).

CRAN landing pages for extRemes and in2extRemes.

1These software packages were funded by the National Science Foundation (NSF) through the NCAR Weather and Climate Impact Assessment Science Initiative with additional support from the NCAR Geophysical Statistics Project (GSP).

2 Originally, extRemes (versions < 2.0) was (primarily) point-and-click software running functions from the R package, ismev. The new package, in2extRemes, now takes on the point-and-click role, and extRemes has only command-line functions. The package ismev is still available on CRAN.

What is EVA

Extreme value analysis (EVA) is used primarily to quantify the stochastic behavior of a process at unusually large (or small) values. Particularly, such analyses usually req uire estimation of the probability of events that are more extreme than any previously observed. Many fields use EVA including: meteorology, hydrology, finance and ocean wave m odeling to name just a few.

Many statistical analyses concern sums or averages of random variables, and often rely upon limiting results such as the Central Limit Theorem to justify use of the normal ( or bell-shaped) distribution. When interest is in extremes, the bulk of the data may be misleading, and the normal distribution is not appropriate. A similar theorem to the Cen tral Limit Theorem, the Extremal Types Theorem, provides justification for using a family of distributions (in the univariate setting, similar results hold for multivariate ana lysis) known as the generalized extreme value (GEV) distribution. This, or analagous results for threshold excesses, are often the focus of EVA.

For more about EVA, see Rick Katz's page on extremes

Back to Top

Getting Started