We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

stat.ME

Change to browse by:

References & Citations

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Statistics > Methodology

Title: Bayesian Prediction with Covariates Subject to Detection Limits

Abstract: Missing values in covariates due to censoring by signal interference or lack of sensitivity in the measuring devices are common in industrial problems. We propose a full Bayesian solution to the prediction problem with an efficient Markov Chain Monte Carlo (MCMC) algorithm that updates all the censored covariate values jointly in a random scan Gibbs sampler. We show that the joint updating of missing covariate values can be at least two orders of magnitude more efficient than univariate updating. This increased efficiency is shown to be crucial for quickly learning the missing covariate values and their uncertainty in a real-time decision making context, in particular when there is substantial correlation in the posterior for the missing values. The approach is evaluated on simulated data and on data from the telecom sector. Our results show that the proposed Bayesian imputation gives substantially more accurate predictions than na\"ive imputation, and that the use of auxiliary variables in the imputation gives additional predictive power.
Subjects: Methodology (stat.ME); Applications (stat.AP)
Cite as: arXiv:2201.07874 [stat.ME]
  (or arXiv:2201.07874v1 [stat.ME] for this version)

Submission history

From: Caroline Svahn [view email]
[v1] Wed, 19 Jan 2022 22:07:43 GMT (226kb)

Link back to: arXiv, form interface, contact.