We humans are constantly trying to protect ourselves from disasters of all kinds. We call events that occur rather rarely but can cause us great harm, extreme events. Fortunately, mathematicians have developed mathematical models that can help us predict the timing and extent of such catastrophic events. In general, we know of two models. The first is the peak-over-threshold (POT) model, which cap ures all events that are greater than the previously defined threshold. The second is the maximum model, which takes into account the extreme events from each of the disjoint, equally long time intervals. Both models lead us to three main distributions of the extreme event theory, the so-called Frechet, Gumbell and Weibull distributions. Using known techniques for predicting parameters of our empirical distribution, we can select one of these distributions as the most appropriate for our data set. So our goal is achieved. We have obtained an analytical function that best represents our data.
|
