J-divergence estimator for scoring models

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F00216224%3A14310%2F12%3A00061863" target="_blank" >RIV/00216224:14310/12:00061863 - isvavai.cz</a>

Výsledek na webu

—

DOI - Digital Object Identifier

—

Jazyk výsledku

angličtina

Název v původním jazyce

J-divergence estimator for scoring models

Popis výsledku v původním jazyce

J-divergence is widely used to describe the difference between two probability distributions $F_{0}$ and $F_{1}$. It is also called the Information value for the purpose of scoring models. Empirical estimate using deciles of scores is the common way howto compute it. However, it may lead to strongly biased results. Moreover, there are some computational issues to solve. To avoid these issues and to lower the bias, the empirical estimate with supervised interval selection (esis) can be used. It is basedon idea of constructing such intervals of scores which ensure to have sufficiently enough observations in each interval. The quantile function $F_{0}^{-1}$ is used for this purpose. For further reduction of the bias and the MSE, new algorithm esis1 wasproposed. The modification lies in the employment of both $F_{0}^{-1}$ and $F_{1}^{-1}$ and the idea of passing data just once with no need to merge the constructed intervals.

Název v anglickém jazyce

J-divergence estimator for scoring models

Popis výsledku anglicky

J-divergence is widely used to describe the difference between two probability distributions $F_{0}$ and $F_{1}$. It is also called the Information value for the purpose of scoring models. Empirical estimate using deciles of scores is the common way howto compute it. However, it may lead to strongly biased results. Moreover, there are some computational issues to solve. To avoid these issues and to lower the bias, the empirical estimate with supervised interval selection (esis) can be used. It is basedon idea of constructing such intervals of scores which ensure to have sufficiently enough observations in each interval. The quantile function $F_{0}^{-1}$ is used for this purpose. For further reduction of the bias and the MSE, new algorithm esis1 wasproposed. The modification lies in the employment of both $F_{0}^{-1}$ and $F_{1}^{-1}$ and the idea of passing data just once with no need to merge the constructed intervals.

Druh

O - Ostatní výsledky

CEP obor

BB - Aplikovaná statistika, operační výzkum

OECD FORD obor

—

Projekt

—

Návaznosti

S - Specificky vyzkum na vysokych skolach

Rok uplatnění

2012

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů