NormalizeBetweenData: Normalization between training and testing sets

In cosmonet-package/COSMONET: Screening-network Cox methods for survival analysis

Description

This function performs the normalization between training and testing set. We normalize the data so that the two sets have the same shape (or distribution).

Usage

NormalizeBetweenData(x1, x2, norm.method = "quantile", plot = FALSE)

Arguments

x1	a matrix of intensities, rows are genes and columns are samples pxn1.
x2	a matrix of intensities, rows are genes and columns are samples pxn2.
plot	logical flag for plot the density distribution. Default is plot=FALSE.
method	character string specifying the quantile normalization method.

Details

This function uses a quantile normalization in order to make the distributions of training and testing set the same across samples. The normalization approach used in our package consist of adding by column each sample of the test set to the train set and normalize the new dataset. Then, we take the test column normalized and build the normalized testing set column by column. This improves the performance and stability of the models and make the two datasets comparable between them.

Value

The following objects are returned:

x1.norm	a normalized matrix n1xp
x2.norm	a normalized matrix n2xp

cosmonet-package/COSMONET documentation built on Dec. 24, 2021, 9:12 p.m.