estimateShuffle: Calculate the shuffle estimators

In shuffle: The Shuffle Estimator for Explainable Variance

Description

estimateShuffle estimates the following quantities for a response vector: the signal variance (signalVar), the noise variance (noiseVar), the total variance (YVar), and the explainable variance (effect). Inputs to the function are the response vector, and a preprocessing structure (the output of prepareShuffle) which holds the design, the shuffle permutation, and the calculated normalizer.

Usage

estimateShuffle(dat, prep, neg = FALSE)

Arguments

dat	A vector of reponses - should be of the same size as the design vector and the shuffle permutation.
prep	The output of prepareShuffle; includes the design, the shuffling permuation, and a normalizer.
neg	If neg=FALSE does not allow the signal variance to get arbitrary negative values, but instead sets signal variance to -1e-05.

Details

estimateShuffle compares the mean-square-between of the data to the mean-square-between of the permuted data, the difference being the scaled noise variance. Effect size is the ratio between the estimated signal data and the estimated total variance.

Value

signalVar	The estimated variance of the signal
noiseVar	The estimated variance of the noise
YVar	The estimated total variance
effect	The proportion of explainable variance (signalVar/Yvar)

Author(s)

Yuval Benjamini

References

Benjamini and Yu (2013), "The shuffle estimator for explainable variance in fMRI experiments".

shuffle documentation built on May 2, 2019, 1:45 p.m.