Rareorextreme events are events that occur with low frequency, and often refers to infrequent events that have a widespread effect and which might destabilize systems (for example, stock markets,[1] ocean wave intensity[2] or optical fibers[3] or society[4]). Rare events encompass natural phenomena (major earthquakes, tsunamis, hurricanes, floods, asteroid impacts, solar flares, etc.), anthropogenic hazards (warfare and related forms of violent conflict, acts of terrorism, industrial accidents, financial and commodity market crashes, etc.), as well as phenomena for which natural and anthropogenic factors interact in complex ways (epidemic disease spread, global warming-related changes in climate and weather, etc.).
Rare or extreme events are discrete occurrences of infrequently observed events. Despite being statistically improbable, such events are plausible insofar as historical instances of the event (or a similar event) have been documented.[5] Scholarly and popular analyses of rare events often focus on those events that could be reasonably expected to have a substantial negative effect on a society—either economically[6] or in terms of human casualties[7] (typically, both). Examples of such events might include an 8.0+ Richter magnitude earthquake, a nuclear incident that kills thousands of people, or a 10%+ single-day change in the value of a stock market index.[8][9][10]
Rare event modeling (REM) refers to efforts to characterize the statistical distribution parameters, generative processes, or dynamics that govern the occurrence of statistically rare events, including but not limited to highly influential natural or human-made catastrophes. Such “modeling” may include a wide range of approaches, including, most notably, statistical models for analyzing historical event data[11][12] and computational software models that attempt to simulate rare event processes and dynamics.[13] REM also encompasses efforts to forecast the occurrence of similar events over some future time horizon, which may be of interest for both scholarly and applied purposes (e.g., risk mitigation and planning).[14] Novel data collection techniques can be used for learning about rare events data.[15]
In many cases, rare and catastrophic events can be regarded as extreme-magnitude instances of more mundane phenomena. For example, seismic activity, stock market fluctuations, and acts of organized violence all occur along a continuum of extremity, with more extreme-magnitude cases being statistically more infrequent.[16] Therefore, rather than viewing rare event data as its own class of information, data concerning "rare" events often exists as a subset of data within a broader parent event class (e.g., a seismic activity data set would include instances of extreme earthquakes, as well as data on much lower-intensity seismic events).
The following is a list of data sets focusing on domains that are of broad scholarly and policy interest, and where "rare" (extreme-magnitude) cases may be of particularly keen interest due to their potentially devastating consequences. Descriptions of the data sets are extracted from the source websites or providers.