1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
|
\name{loessByCol}
\alias{loessByCol}
\alias{locfitByCol}
\title{Locally Weighted Mean By Column}
\description{Smooth columns of matrix by non-robust loess curves of degree 0.}
\usage{
loessByCol(y, x=NULL, span=0.5)
locfitByCol(y, x=NULL, weights=1, span=0.5, degree=0)
}
\arguments{
\item{y}{numeric matrix of response variables.}
\item{x}{numeric covariate vector of length \code{nrow(y)}, defaults to equally spaced.}
\item{span}{width of the smoothing window, in terms of proportion of the data set. Larger values produce smoother curves.}
\item{weights}{relative weights of each observation, one for each covariate value.}
\item{degree}{degree of local polynomial fit}
}
\value{A list containing a numeric matrix with smoothed columns and a vector of leverages for each covariate value.
\code{locfitByCol} returns a numeric matrix.
}
\details{
Fits a loess curve with degree 0 to each column of the response matrix, using the same covariate vector for each column.
The smoothed column values are tricube-weighted means of the original values.
\code{locfitByCol} uses the \code{locfit.raw} function of the \code{locfit} package.
}
\author{Aaron Lun for \code{loessByCol}, replacing earlier R code by Davis McCarthy. Gordon Smyth for \code{locfitByCol}.}
\seealso{
\code{\link{loess}}
}
\examples{
y <- matrix(rnorm(100*3), nrow=100, ncol=3)
head(y)
out <- loessByCol(y)
head(out$fitted.values)
}
|
