Extreme value analysis: a better way to analyse rare datasets / Zuraida Jaafar

Imagine being in the centre of Malaysia, amidst the vibrant cities, lush rainforests, and breath-taking scenery that make up this lovely country. The weather in Malaysia often has sunny days, occasional rain showers, and light breezes. This predictable weather is what you would consider the "normal" situation. However, once in a blue moon, a powerful monsoon passes through a few locations especially, on the shores of the South China Sea with heavy rain and strong winds. Many places are in danger of being submerged by rivers that overflow and the sea that surges. This extraordinary weather events represents extreme value data. Extreme value data refers to observations or data points that represent exceptionally rare or extreme events within a dataset. These events are situated in the tails of the probability distribution and are typically characterized by their infrequent occurrence and significant deviation from the typical or expected values. Extreme value data can be found in various fields, from finance (market crashes) to meteorology (severe storms), and from healthcare (rare diseases) to engineering (catastrophic failures).