Koriģētā empīriskās ticamības funkcija kvantilēm
Loading...
Date
Authors
Advisor
Journal Title
Journal ISSN
Volume Title
Publisher
Latvijas Universitāte
Language
lav
Abstract
Kopš Owen iepazīstināja ar empīriskās ticamības metodi (EL), tā kļuvusi par plaši pielietojamu rīku, lai novērtētu parametrus datu kopām ar grūti nosakāmiem sadalījumiem. 2008.gadā tika ieviests empīriskās ticamības metodes paplašinājums - koriģētā empīriskās ticamības metode (AEL). Darbā veikts abu metožu salīdzinājums vidējās vērtības un kvantiles novērtēšanai, ieskaitot to gludo ekvivalentu izpēti (SEL un SAEL). Darbā tiek aprakstīts teorētiskais pamatojums, t.sk., pamata pieņēmumi, statistiskās īpašības, kā arī apstākļi, kādos katra metode ir ieteicama. Šī pētījuma pamatā ir rūpīgi izstrādātu simulāciju sērija dažādiem sadalījuma datiem. Simulācijas veiktas, lai demonstrētu EL, AEL, SEL un SAEL veiktspēju, konstruējot ticamības intervālus.
Since Owen introduced the empirical likelihood (EL) method, it has become a widely used tool for estimating parameters for datasets with hard-to-detect distributions. In 2008, an extension of the empirical likelihood method, the adjusted empirical likelihood (AEL) method, was introduced.Two methods for estimating the mean and quantile, including an examination of their smooth counterparts (SEL and SAEL) are compared in the thesis. The theoretical background, including the underlying assumptions, statistical properties and the circumstances under which each method is recommended, is described. This study is based on a series of carefully designed simulations for a variety of distributional data. The simulations are performed to demonstrate the performance of EL, AEL, SEL and SAEL in constructing confidence intervals.
Since Owen introduced the empirical likelihood (EL) method, it has become a widely used tool for estimating parameters for datasets with hard-to-detect distributions. In 2008, an extension of the empirical likelihood method, the adjusted empirical likelihood (AEL) method, was introduced.Two methods for estimating the mean and quantile, including an examination of their smooth counterparts (SEL and SAEL) are compared in the thesis. The theoretical background, including the underlying assumptions, statistical properties and the circumstances under which each method is recommended, is described. This study is based on a series of carefully designed simulations for a variety of distributional data. The simulations are performed to demonstrate the performance of EL, AEL, SEL and SAEL in constructing confidence intervals.