Plotting ts objects

This document explains time series related plotting using ggplot2 and {ggfortify}.

Plotting ts objects

{ggfortify} let {ggplot2} know how to interpret ts objects. After loading {ggfortify}, you can use ggplot2::autoplot function for ts objects.


plot of chunk unnamed-chunk-1

To change line colour and line type, use ts.colour and ts.linetype options. Use help(autoplot.ts) (or help(autoplot.*) for any other objects) to check available options.

autoplot(AirPassengers, ts.colour = 'red', ts.linetype = 'dashed')

plot of chunk unnamed-chunk-2

Multivariate time series will be drawn with facets.


plot of chunk unnamed-chunk-3

Specify facets = FALSE to draw on single axes.

autoplot(Canada, facets = FALSE)

plot of chunk unnamed-chunk-4

Also, autoplot can handle other time-series-likes. Supported packages are:

autoplot(as.xts(AirPassengers), ts.colour = 'green')

autoplot(as.timeSeries(AirPassengers), ts.colour = ('dodgerblue3'))

Specifying geometrics

You can change {ggplot2} geometrics specifying by its name. Geometrics currently supported are line, bar, ribbon and point.

autoplot(AirPassengers, ts.geom = 'bar', fill = 'blue')

plot of chunk unnamed-chunk-6

autoplot(AirPassengers, ts.geom = 'ribbon', fill = 'green')

plot of chunk unnamed-chunk-7

autoplot(AirPassengers, ts.geom = 'point', shape = 3)

plot of chunk unnamed-chunk-8

As described above, multivariate timeseries can be drawn in a single grid specifying facets = FALSE. Time series are not stacked by default. Specifying stacked = TRUE allows stacking.

mts <- ts(data.frame(a = c(1, 2, 3, 4, 4, 3), b = c(3, 2, 3, 2, 2, 1)), start = 2010)
autoplot(mts, ts.geom = 'bar', facets = FALSE)

plot of chunk unnamed-chunk-9

autoplot(mts, ts.geom = 'bar', facets = FALSE, stacked = TRUE)

plot of chunk unnamed-chunk-10

autoplot(mts, ts.geom = 'ribbon', facets = FALSE)

plot of chunk unnamed-chunk-11

autoplot(mts, ts.geom = 'ribbon', facets = FALSE, stacked = TRUE)

plot of chunk unnamed-chunk-12

Plotting with forecast package

{ggfortify} supports forecast object in the {forecast} package.

d.arima <- auto.arima(AirPassengers)
d.forecast <- forecast(d.arima, level = c(95), h = 50)

plot of chunk unnamed-chunk-13

There are some options to change basic settings.

autoplot(d.forecast, ts.colour = 'firebrick1', predict.colour = 'red',
         predict.linetype = 'dashed', = FALSE)

plot of chunk unnamed-chunk-14

Plotting with vars package

ggfortify supports varpred object in vars package.

d.vselect <- VARselect(Canada, lag.max = 5, type = 'const')$selection[1]
d.var <- VAR(Canada, p = d.vselect, type = 'const')

Available options are the same as forecast.

autoplot(predict(d.var, n.ahead = 50), ts.colour = 'dodgerblue4',
         predict.colour = 'blue', predict.linetype = 'dashed')

plot of chunk unnamed-chunk-16

Plotting with changepoint package

{ggfortify, eval=hasDep} supports cpt object in {changepoint} package.


plot of chunk unnamed-chunk-17

You can change some options for cpt.

autoplot(cpt.meanvar(AirPassengers), cpt.colour = 'blue', cpt.linetype = 'solid')

plot of chunk unnamed-chunk-18

Plotting with strucchange package

ggfortify supports breakpoints object in strucchange package. Same plotting options as changepoint are available.

autoplot(breakpoints(Nile ~ 1), ts.colour = 'blue', ts.linetype = 'dashed',
         cpt.colour = 'dodgerblue3', cpt.linetype = 'solid')

plot of chunk unnamed-chunk-19

Plotting with {KFAS} package

You can use autoplot in almost the same manner as {dlm}. Note that autoplot draws smoothed result if it exists in KFAS::KFS instance, and KFAS::KFS contains smoothed result by default.

model <- SSModel(
  Nile ~ SSMtrend(degree=1, Q=matrix(NA)), H=matrix(NA)
fit <- fitSSM(model=model, inits=c(log(var(Nile)),log(var(Nile))), method="BFGS")
smoothed <- KFS(fit$model)

plot of chunk unnamed-chunk-20

If you want filtered result, specify smoothing='none' when calling KFS. For details, see help(KFS).

filtered <- KFS(fit$model, filtering="mean", smoothing='none')

plot of chunk unnamed-chunk-21

Also, KFAS::signal will retrieve specific state from KFAS::KFS instance. The result will be a list which contains the retrieved state as ts in signal attribute. ggfortify can autoplot it using class inference.

trend <- signal(smoothed, states="trend")
## [1] "list"

Because signal is a ts instance, you can use autoplot and p option as the same as dlm::dlmSmooth example.

p <- autoplot(filtered)
autoplot(trend, ts.colour = 'blue', p = p)

plot of chunk unnamed-chunk-23

Plotting time series statistics

{ggfortify} supports following time series related statistics in stats package:

autoplot(stl(AirPassengers, s.window = 'periodic'), ts.colour = 'blue')

plot of chunk unnamed-chunk-24

NOTE With acf and spec.*, specify plot = FALSE to suppress default plotting outputs.

autoplot(acf(AirPassengers, plot = FALSE))

plot of chunk unnamed-chunk-25

You can pass some options when plotting acf via autoplot. Built-in acf calcurates the confidence interval at plotting time and doesn’t hold the result, equivalent options can be passed to autoplot. Following example shows to change the value of confidence interval and method (use ma assuming the input follows MA model).

autoplot(acf(AirPassengers, plot = FALSE), = '#0000FF', = 0.8, = 'ma')

plot of chunk unnamed-chunk-26

autoplot(, plot = FALSE))

plot of chunk unnamed-chunk-27

ggcpgram should output the cumulative periodogram as the same as cpgram. Because cpgram doesn’t have return value, we cannot use autoplot(cpgram(...)).

ggcpgram(arima.sim(list(ar = c(0.7, -0.5)), n = 50))

plot of chunk unnamed-chunk-28

ggtsdiag should output the similar diagram as tsdiag.


plot of chunk unnamed-chunk-29

ggfreqplot is a genelarized month.plot. You can pass freq if you want, otherwise time-series’s frequency will be used.


plot of chunk unnamed-chunk-30

ggfreqplot(AirPassengers, freq = 4)

plot of chunk unnamed-chunk-30