CMStatistics 2022: Start Registration
View Submission - CMStatistics
B1877
Title: Time series prediction and extreme events Authors:  Manuela Neves - FCiencias.ID, Associacao para a Investigacao e Desenvolvimento de Ciencias (Portugal) [presenting]
Clara Cordeiro - FCiencias.ID, Associacao para a Investigacao e Desenvolvimento de Ciencias (Portugal) (Portugal)
Abstract: Time series forecasting, i.e. making predictions from historical data available, is of major importance with a wide variety of applications, such as in finance, weather, the healthcare sector, etc. It is an intensively studied topic, but the existence of extreme events can result in weak performance and low accuracy in the results. Extreme events are rare but do play a critical role in many real applications. Whenever the focus is on large values, estimation is usually performed based on the largest $k$ order statistics in the sample or on the excesses over a high level $u$. In Extreme Value Analysis and whenever dealing with large values, a few primordial parameters need to be adequately estimated. As we are interested in forecasting extremes in time series, procedures on time series and extreme value theory will come together. Resampling techniques and time series methods for modelling and predicting a time series are computational procedures proposed to improve the performance of the results. A simulation study and applications to real data sets are performed in the R software.