Bayes Multiple Decision Functions

TL;DR

This paper develops a Bayesian decision-theoretic framework for multiple binary decision-making with dependent data, introducing the Bayes multiple decision function (BMDF) and efficient algorithms, applicable to high-dimensional and failure-time data.

Contribution

It extends existing methods by incorporating dependent data structures via frailty-induced copulas and provides a practical algorithm for optimal decisions in complex, high-dimensional settings.

Findings

The proposed method effectively controls false discovery and missed discovery rates.

Simulation studies demonstrate the approach's accuracy and efficiency.

Application to microarray data shows practical utility in real-world biological data.

Abstract

This paper deals with the problem of simultaneously making many (M) binary decisions based on one realization of a random data matrix X. M is typically large and X will usually have M rows associated with each of the M decisions to make, but for each row the data may be low dimensional. A Bayesian decision-theoretic approach for this problem is implemented with the overall loss function being a cost-weighted linear combination of Type I and Type II loss functions. The class of loss functions considered allows for the use of the false discovery rate (FDR), false nondiscovery rate (FNR), and missed discovery rate (MDR) in assessing the decision. Through this Bayesian paradigm, the Bayes multiple decision function (BMDF) is derived and an efficient algorithm to obtain the optimal Bayes action is described. In contrast to many works in the literature where the rows of the matrix X are assumed to be stochastically independent, we allow in this paper a dependent data structure with the associations obtained through a class of frailty-induced Archimedean copulas. In particular, non-Gaussian dependent data structure, which is the norm rather than the exception when dealing with failure-time data, can be entertained. The numerical implementation of the determination of the Bayes optimal action is facilitated through sequential Monte Carlo techniques. The main theory developed could also be extended to the problem of multiple hypotheses testing, multiple classification and prediction, and high-dimensional variable selection. The proposed procedure is illustrated for the simple versus simple and for the composite hypotheses setting via simulation studies. The procedure is also applied to a subset of a real microarray data set from a colon cancer study.