EXTREME VALUE THEORY
| Collection | International Encyclopedia of Systems and Cybernetics |
|---|---|
| Year | 2004 |
| Vol. (num.) | 2(1) |
| ID | ◀ 1235 ▶ |
| Object type | Methodology or model |
A forecasting set of models that can be used to predict the likehood of extremely rare events .
The theory is based on distribution curves (Gumbel, Fréchet and Weibull curves) which give “a precise quantitative statement of the otherwise vague notion that the more extreme the event, the less likely it is to happen” (R. MATTHEWS, 1996, p. 37-38).
The theory has been developed by R. SMITH (Univ. of North Carolina) and L.DE HAAN (Erasmus University, Rotterdam).
It offers no real possibility of precise predictions but is useful for determining the most extreme possible events in accordance with statistics of lesser similar events in the past, on a long time scale .
Dutch experts have found for example that the maximum known historical flood due to a North Sea storm in Holland has topped 4 meters above sea level (in 1570). It has thus been estimated that walls of 5 meters upon the sea level should meet a criterion of a once in
10 000 years flood. (see GOLDENFELD and KADANOFF, 1999, p. 87)
Other examples are volcanic eruptions, big earthquakes, crashes in stock markets, epidemics in populations , hurricanes in the Caribean zone, the “El Ni o” events in the Pacific Ocean. Intermittence generally implies sudden paroxystic changes or an epochal reversa l in-between variable but long relaxation periods . It is related to synergetic processes, power laws , and dissipative structuration , all of these probably governed by great fluctuations of energy inputs , or their cumulative long term effects (Ibid, p.87-89).
See also
Numbers (Prime), Periodicity, Term (short-, medium-, long-)