tricubeMovingAverage {limma}R Documentation

Moving Average Smoother With Tricube Weights

Description

Apply a moving average smoother with tricube distance weights to the columns of a matrix.

Usage

tricubeMovingAverage(x,span=0.5,full.length=TRUE)

Arguments

x

numeric vector

span

proportion of points included in the local window

full.length

logical value, should output have same number of length as input?

Details

This function smooths a vector (considered as a time series) using a moving average with tricube weights. This is similar to a loess curve of degree zero, with a couple of differences: a continuity correction is applied when computing the neighbouring points and, when full.length=TRUE, the span halves at the end points.

The filter function in the stats package is called to do the low-level calculations.

This function is used by barcodeplot to compute enrichment worms.

Value

Numeric vector of smoothed values. If full.length=TRUE, of same length as x. If full.length=FALSE, has width-1 fewer rows than x.

Author(s)

Gordon Smyth

See Also

filter, barcodeplot

Examples

x <- rbinom(100,size=1,prob=0.5)
plot(1:100,tricubeMovingAverage(x))

[Package limma version 3.24.14 Index]