A1251
Title: A tail-based estimation approach for the shape parameter of the three-parameter log-logistic distribution
Authors: Frederico Caeiro - NOVA.ID.FCT - Universidade Nova de Lisboa (Portugal) [presenting]
Abstract: The log-logistic distribution is a flexible heavy-tailed model, widely applied in fields such as survival analysis, hydrology, insurance, economics, and finance. This distribution is commonly presented with two parameters: a shape parameter and a scale parameter. To increase flexibility, an additional location parameter can be added, resulting in the three-parameter log-logistic distribution. A new estimator for the shape parameter of the three-parameter log-logistic distribution that reduces bias is introduced. This estimator is based on the largest observations from the sample and can be used in conjunction with existing best linear unbiased estimators for the location and scale parameters. Various properties of the proposed estimator are derived. Additionally, a simple threshold selection procedure for the tail sample fraction is presented.
