Closed ellisp closed 8 years ago
Basic--fairly horrific--plotting is implemented here: 657c5b83a967f59ee0668ffaffbe3e9f6bb01f84
It merely plots the original series and the fitted values. The idea of plotting each of the models by setting mfrow
won't work because some of the individual models themselves change this in their own plot methods. Upon reflection, the "Hit plot.lm()
is an anomaly and shouldn't be emulated. We could just splatter all of the individual model plots one after another.
This second behavior is implemented here with type = "models"
: 241e002e28b08070eda16533cb5d5af774650f26
We can add ggplot support and other plot types later, especially once the forecast package itself implemented ggplot.
Any ideas for plotting our ensemble object? Want to include ggplot2 in addition to base graphics? For the individual component models we can use the individual plot methods from the "forecast" package for each model. For example, to see the next plot:" prompt like in the
plot(hybridModel(wineind, models = "aten"))
could be setup to create a 2x2 plot bypar(mfrow = c(2, 2))
and then plot the Arima/tbats/ets/nnetar objects using the forecast package's methods. The nnetar object does not have a plot method, so that will need to be excluded, and theplot.Arima()
method already setspar(mfrow = c(1, 2))
, so this approach might need some work. If we want to plot both the individual component models and something for the ensemble we could put in a "Hitplot.lm()
method