rfbaseline: Robust Fitting of Baselines

Description

Robust fitting of local regression models for estimating a baseline or a background signal

Usage

rfbaseline(x, y, span = 2/3, NoXP = NULL,
           maxit = c(2, 2), b = 3.5, weight = NULL,
           Scale = function(r) median(abs(r))/0.6745,
           delta = NULL, SORT = TRUE, DOT = FALSE, init = NULL)

Arguments

x, y

Abscissa and ordinate of the points on the scatterplot.

span

Specifies the amount of smoothing; span is the fraction of points used to compute each fitted value; as span increases the output becomes smoother.

NoXP

Another way of specifying the amount of smoothing; NoXP is the Number of X Points used to compute each fitted value; it must be larger than 3.

maxit

The number of iterations in the robust fit; if maxit=c(0,0), the nonrobust fit is returned; the first entry specifies the number of iterations using an asymmetric biweight function, whereas the second entry specifies the number of ite

Tuning constant in the biweight function.

weight

Optional weights to be given to individual observations.

Scale

function specifying how to calculate the scale of the residuals.

delta

Nonnegative parameter which may be used to save computation. By default, if

length(x) <= 100<="" code="">, delta is
    set equal to 0; if length(x) > 100 set to 1/100th of the
    range of x.

SORT

Boolean variable indicating whether x data must be sorted. Change it only when the x are sorted and you want to safe computer time.

DOT

If TRUE disregard outliers totally; that is, observations with weight 0 are disregarded even when the neighbourhood is determined.

init

Values of an initial fit.

Value

List containing components
xSorted input vector x with duplicate points removed
yCorresponding input vector y
fitFitted values at x
rwRobust weights of (x,y)-Points used in last iteration of fit
scaleScale used in last iteration of fit

References

Ruckstuhl, Andreas F., Matthew P. Jacobson, Robert W. Field and James A. Dodd (2001); Baseline Subtraction Using Robust Local Regression Estimation; Journal of Quantitative Spectroscopy and Radiative Transfer 68: 179 -- 193 Ruckstuhl, Andreas F., et al.; Estimation of background concentrations of atmospheric trace gases using robust local regression; to be published

Examples

Run this code

data(MS)
MS1 <- log10(MS[MS$mz>12000&MS$mz<1e5,])

MS1.rfb2 <- rfbaseline(x=MS1$mz, y=MS1$I, NoXP=2200, maxit=c(5,0))
plot(x=MS1$mz, y=MS1$I,ty="l",
     xlab="log(mass/charge)", ylab="log(intensity)")
lines(MS1.rfb2$x, MS1.rfb2$fit, col="orange", lwd=3)

MS1.rfb3 <- rfbaseline(x=MS1$mz, y=MS1$I, NoXP=1100, maxit=c(5,0),
                       DOT=TRUE, Scale=function(x) mad(x, center=0))
plot(x=MS1$mz, y=MS1$I,ty="l",
     xlab="log(mass/charge)", ylab="log(intensity)")
lines(MS1.rfb3$x, MS1.rfb3$fit, col="orange", lwd=3)

## 'delta=0' needs much more computer time
MS1.rfb4 <- rfbaseline(x=MS1$mz, y=MS1$I, NoXP=2200,
                       delta=0, maxit=c(5,0))
plot(x=MS1$mz, y=MS1$I,ty="l",
     xlab="log(mass/charge)", ylab="log(intensity)")
lines(MS1.rfb4$x, MS1.rfb4$fit, col="orange", lwd=3)