Time series smoothing with multiple bandwidth levels

Plots a time series and overlays a family of smoothed curves using a chosen smoothing method. For bandwidth-based methods ("kernel", "lowess", "splines", "default"), a sequence of bandwidth or span values from start to stop is swept and each curve is drawn in a different rainbow colour, making it easy to visually select an appropriate smoothing level. "polynomial" and "linear" ignore the bandwidth sequence and fit regression-based trend lines instead. The function uses base R graphics and produces a plot as a side effect.

Usage

ts_smoothing(
  df,
  start = 0.01,
  stop = 2,
  step = 0.001,
  title = "",
  type = "kernel"
)

Arguments

df

A ts object containing the time series to smooth.

start

Numeric. Starting value of the bandwidth or span sequence. Default is 0.01.

stop

Numeric. Ending value of the bandwidth or span sequence. Default is 2.

step

Numeric. Increment between bandwidth values. Smaller values produce more curves. Default is 0.001.

title

Character string appended to the plot title. Default is "".

type

Character string specifying the smoothing method. One of:

"kernel"

Gaussian kernel smoother via ksmooth() — bandwidth controls the smoothing window (default).

"lowess"

Locally weighted regression via lowess() — f controls the span proportion.

"friedman"

Friedman's super-smoother via supsmu() — span must be in (0, 1).

"splines"

Smoothing splines via smooth.spline(), where spar controls the penalty.

"default"

Running mean filter via filter() — bandwidth rounded to an integer window width.

"polynomial"

Fits a centred cubic polynomial trend with and without seasonal (cos / sin) terms via lm().

"linear"

Fits a simple linear trend via lm() and draws the regression line.

Value

Invisibly returns NULL. The function is called for its side effect of producing a base R plot.

Examples

ts_data <- ts(UKDriverDeaths, start = 1969, end = 1984, frequency = 12)
par(mfrow = c(2, 2))
ts_smoothing(ts_data,
  start = .01, stop = 2, step = .01,
  title = "Driver Deaths in UK", type = "default"
)
ts_smoothing(ts_data,
  start = .01, stop = 2, step = .01,
  title = "Driver Deaths in UK", type = "polynomial"
)
ts_smoothing(ts_data,
  start = .01, stop = 2, step = .01,
  title = "Driver Deaths in UK", type = "linear"
)
ts_smoothing(ts_data,
  start = .01, stop = 2, step = .01,
  title = "Driver Deaths in UK", type = "kernel"
)

ts_smoothing(ts_data,
  start = .01, stop = 2, step = .01,
  title = "Driver Deaths in UK", type = "lowess"
)
ts_smoothing(ts_data,
  start = .01, stop = 2, step = .01,
  title = "Driver Deaths in UK", type = "friedman"
)
ts_smoothing(ts_data,
  start = .01, stop = 2, step = .01,
  title = "Driver Deaths in UK", type = "splines"
)