Random measures are the key ingredient for effective nonparametric Bayesian modeling of time-to-event data. This paper focuses on priors for the hazard rate function, a popular choice being the kernel mixture with respect to a gamma random measure. Sampling schemes are usually based on approximations of the underlying random measure, both a priori and conditionally on the data. Our main goal is the quantification of approximation errors through the Wasserstein distance. Though easy to simulate, the Wasserstein distance is generally difficult to evaluate, making tractable and informative bounds essential. Here we accomplish this task on the wider class of completely random measures, yielding a measure of discrepancy between many noteworthy random measures, including the gamma, generalized gamma and beta families. By specializing these results to gamma kernel mixtures, we achieve upper and lower bounds for the Wasserstein distance between hazard rates, cumulative hazard rates and survival functions.

Approximation of Bayesian models for time-to-event data / Catalano, Marta; Lijoi, Antonio; Pruenster, Igor. - In: ELECTRONIC JOURNAL OF STATISTICS. - ISSN 1935-7524. - 14:2(2020), pp. 3366-3395. [10.1214/20-EJS1746]

Approximation of Bayesian models for time-to-event data

Catalano, Marta;
2020

Abstract

Random measures are the key ingredient for effective nonparametric Bayesian modeling of time-to-event data. This paper focuses on priors for the hazard rate function, a popular choice being the kernel mixture with respect to a gamma random measure. Sampling schemes are usually based on approximations of the underlying random measure, both a priori and conditionally on the data. Our main goal is the quantification of approximation errors through the Wasserstein distance. Though easy to simulate, the Wasserstein distance is generally difficult to evaluate, making tractable and informative bounds essential. Here we accomplish this task on the wider class of completely random measures, yielding a measure of discrepancy between many noteworthy random measures, including the gamma, generalized gamma and beta families. By specializing these results to gamma kernel mixtures, we achieve upper and lower bounds for the Wasserstein distance between hazard rates, cumulative hazard rates and survival functions.
2020
Bayesian nonparametrics, completely random measures, gamma random measure, kernel mixtures, posterior sampling, Survival analysis, Wasserstein distance
Approximation of Bayesian models for time-to-event data / Catalano, Marta; Lijoi, Antonio; Pruenster, Igor. - In: ELECTRONIC JOURNAL OF STATISTICS. - ISSN 1935-7524. - 14:2(2020), pp. 3366-3395. [10.1214/20-EJS1746]
File in questo prodotto:
File Dimensione Formato  
EJS_2020.pdf

Open Access

Tipologia: Versione dell'editore
Licenza: Creative commons
Dimensione 565.3 kB
Formato Adobe PDF
565.3 kB Adobe PDF Visualizza/Apri
Pubblicazioni consigliate

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11385/232220
Citazioni
  • Scopus 3
  • ???jsp.display-item.citation.isi??? 2
social impact