Publication

Abstract : Extreme value theory is a statistical method for modeling and evaluating risks in peculiar situations that has gained popularity in risk management. The main emphasis is on the tail behavior of the underlying distributions. Threshold selection techniques have been extensively used to estimate the extreme quantiles or tails of the distribution. An optimal threshold selection has been an unresolved problem in the peaks over threshold method of this theory for several decades. Exceedances above a certain threshold typically follow the generalized Pareto distribution asymptotically. Our primary goal is to create an efficient threshold selection technique that can be modeled with the appropriate extreme value distribution to evaluate the sample of exceedances above an appropriate threshold. In this paper, we present a pragmatic automated method for threshold selection based on the distribution of parameter estimates and evaluation indicator value from entropy-based weighted multiple testing concepts. A simulation study has been performed to assess its effectiveness, and the suggested method has been demonstrated through the monthly observation of ammonium ion concentration in the river water dataset.