onlineFDR-package {onlineFDR}R Documentation

onlineFDR: A package for online FDR control

Description

The onlineFDR package provides methods to control the false discovery rate (FDR) for online hypothesis testing, where hypotheses arrive sequentially in a stream. A null hypothesis is rejected based only on the previous decisions, as the future p-values and the number of hypotheses to be tested are unknown.

Details

Package: onlineFDR
Type: Package
Version: 0.99.7
Date: 2018-05-30
License: GPL-3

Javanmard and Montanari (2015, 2018) proposed two methods for online FDR control. The first is LORD, which stands for (significance) Levels based On Recent Discovery and is implemented by the function LORD. LORDdep provides a modified version of LORD that is valid for dependent p-values.

The second method is LOND, which stands for (significance) Levels based On Number of Discoveries and is implemented by the function LOND. By specifying dep = TRUE, thus function runs a modified version of LOND which is valid for dependent p-values.

As an alternative to these methods, a Bonferroni-like test is implemented by the function bonfInfinite. This procedure is also valid for dependent p-values.

Further details on all these procedures can be found in Javanmard and Montanari (2015, 2018).

Author(s)

David Robertson (david.robertson@mrc-bsu.cam.ac.uk), Adel Javanmard, Andrea Montanari and Natasha Karp.

References

Javanmard, A. and Montanari, A. (2015) On Online Control of False Discovery Rate. arXiv preprint, https://arxiv.org/abs/1502.06197

Javanmard, A. and Montanari, A. (2018) Online Rules for Control of False Discovery Rate and False Discovery Exceedance. Annals of Statistics, 46(2):526-554.


[Package onlineFDR version 1.0.0 Index]