Read First

Disclaimer

pyextremes was created to make the process of running EVA simpler and faster. While the project is built with reasonable defaults which produce good results in most cases, one should not rely on the model as a source of ground truth. It is always the responsibility of the user to understand the subject of analysis and to properly interpret the model outputs.

Example

A 1000-year wave height of 100 meters is not physical and is an artifact of the underlying statistical model. One should always be mindful of the actual process being modeled and remember that the model gives a proabilistic estimate of extreme values under the assumption that the selected model (e.g. GEVD or GPD) correctly describes the underlying process (in this example, water waves).

Tutorial Structure

Each tutorial section covers a particular area of EVA, such as extreme value extraction, fitting a model, or summarizing and visualizing analysis results. pyextremes was built in a modular fashion where each of these components is implemented independently and can be used on its own. In order to make life easier a helper class EVA was created (located in pyextremes.eva.EVA) which chains these components together to streamline the most common types of EVA workflows and to reduce the amount of code a user needs to write when performing analysis.

When possible, sections of this tutorial present two alternative ways to perform the same action: via EVA and via low-level functions which are using by EVA behind the scenes.