winsorizeMatrix: Winsorizing of an entire matrix

In pgpmartin/GeneNeighborhood: Analyze the Neighborhood (Upstream/Downstream Neighbors) of Genes

Description

Winsorizing of an entire matrix. Here, winsorization consists in replacing extreme values in a matrix by the corresponding upper and lower quantiles.

Usage

winsorizeMatrix(mat, low = 1, high = 99, verbose = FALSE)

Arguments

mat	A matrix to winsorize
low	Numeric. A value in [0,100]. All the values below quantile(mat,low/100) are replaced by quantile(mat,low/100)
high	Numeric. A value in [0,100]. All the values above quantile(mat,high/100) are replaced by quantile(mat,high/100)
verbose	Logical. Default to FALSE. If verbose=TRUE, the function will print statistics on the number of values replaced.

Value

A matrix with extreme values replaced by the corresponding quantiles.

Examples

## Create a matrix with random number from N(0,1):
  set.seed(123)
  x <- matrix(rnorm(1e4), ncol=100)
## Replace the top and bottom 1% extreme values by the corresponding quantiles:
  y <- winsorizeMatrix(x, low=1, high=99, verbose = TRUE)
## Add some NA in x:
  x[x>2] <- NA
## winsorizeMatrix still works
  y <- winsorizeMatrix(x, low=1, high=99, verbose = TRUE)

pgpmartin/GeneNeighborhood documentation built on Sept. 2, 2021, 6:37 a.m.