lowess
Scatter Plot Smoothing
Description
Gives a robust, local smooth of scatter plot data.
Among other options is the fraction of data smoothed at each point.
Usage
lowess(x, y = NULL, f = 2/3, iter = 3, delta = 0.01 * diff(range(xy$x[o])))
Arguments
| x,y |
vectors of data for scatter plot.
y can be NULL, then x can be
a univariate time series,
a matrix or data.frame with two columns
or a list with two components named x and y..
|
| f |
fraction of the data used for smoothing at each x point.
The larger the f value, the smoother the fit.
|
| iter |
number of iterations used in computing robust estimates.
|
| delta |
interval size (in units corresponding to x).
If lowess estimates at two x values within delta
of one another,
it fits any points between them by linear interpolation.
The default is 1% of the lagged differences of range of x
coordinates (the first column of x if it is a data structure).
If delta=0 all but identical x values are estimated independently.
|
Details
This is a scatter plot smoother - it does not make any assumptions about the
x values being evenly spaced.
Lowess uses robust locally linear fits.
A window, dependent on f, is placed about each x value;
points that are inside the window
are weighted so that nearby points get the most weight.
Value
list containing components named x and y which
are the x,y points of the smoothed scatter plot.
Note that x is a sorted version of the input x vector,
with duplicate points removed.
Note
This function may be slow for large numbers of points; execution time
is proportional to (iter*f*n^2).
Increasing delta should speed things up, as will decreasing f.
References
Cleveland, W. S. (1979).
Robust locally weighted regression and smoothing scatterplots.
Journal of the American Statistical Association
74, 829-836.
Chambers, J. M., Cleveland, W. S., Kleiner, B. and Tukey, P. A. (1983).
Graphical Methods for Data Analysis.
Wadsworth, Belmont, California.
Becker, R. A., Chambers, J. M. and Wilks, A. R. (1988)
The New S Language.
Wadsworth & Brooks/Cole.
Cleveland, W. S. (1981)
LOWESS: A program for smoothing scatterplots by robust locally weighted
regression. The American Statistician, 35, 54.
See Also
Examples
require(Sdatasets)
fit <- lowess(car.miles, car.gals)
fit2 <- lowess(car.miles, car.gals, f = 0.2)
resid <- car.gals - approx(fit, xout = car.miles)$y
# Same values with a data.frame:
car.df <- data.frame(miles=car.miles, gals=car.gals)
fit <- lowess(car.df)
fit2 <- lowess(car.df, f = 0.8)