pime.oob.error: Baseline noise detection

In microEcology/pime: Prevalence Interval for Microbiome Evaluation

Description

This function builds random forests for sample classification measuring the prediction error of random forests. It wraps on ranger, taking as input a prevalence unfiltered dataset (the original dataset). The model performance is indicated by the out-of-bag (OOB) error rate. Higher OOB error indicates the dataset has a high relative abundance of taxa with low prevalence, which is defined as noise in PIME analysis. There is no formal criteria for definition of low or high OOB error, but empirical tests showed that PIME can improve microbiome differences when OOB error >= 0.01.

Usage

pime.oob.error(physeq, variable)

Arguments

physeq	The input file in phyloseq object format
variable	Any variable present in the metadata to be analyzed. "variable to run the classification"

See Also

ranger

Examples

pime.oob.error(restroom, "Environment")

microEcology/pime documentation built on Nov. 13, 2019, 11:16 p.m.