Discrete power distributions are proposed and studied, by considering the positive jumps on the discontinuities of an original discrete distribution function. Inequalities in moments and distribution functions are studied, allowing the definition of discrete intermediate distributions that lie between an original distribution and a power distribution. Original uniform, binomial, Poisson, negative binomial, and hypergeometric distributions are considered, to propose new power and intermediate distributions. Stochastic orders and unimodality are discussed. Estimation problems using likelihood are investigated. Simultion experiments are performed, to evaluate the bias and the mean square error of the maximum likelihood estimates, that are numrically calculated, with classic tools for numerical optimization.
Discrete power distributions and inference using likelihood
ANDREA PALLINI
2018-01-01
Abstract
Discrete power distributions are proposed and studied, by considering the positive jumps on the discontinuities of an original discrete distribution function. Inequalities in moments and distribution functions are studied, allowing the definition of discrete intermediate distributions that lie between an original distribution and a power distribution. Original uniform, binomial, Poisson, negative binomial, and hypergeometric distributions are considered, to propose new power and intermediate distributions. Stochastic orders and unimodality are discussed. Estimation problems using likelihood are investigated. Simultion experiments are performed, to evaluate the bias and the mean square error of the maximum likelihood estimates, that are numrically calculated, with classic tools for numerical optimization.File | Dimensione | Formato | |
---|---|---|---|
8598-29799-1-PB.pdf
accesso aperto
Tipologia:
Versione finale editoriale
Licenza:
Creative commons
Dimensione
415.46 kB
Formato
Adobe PDF
|
415.46 kB | Adobe PDF | Visualizza/Apri |
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.