plot.statespace displays different graphs expressing the behavior of the state-space
model:
1. Evolution of the Bias: rolling version of the error of the model.
2. Evolution of the RMSE: root-mean-square-error computed on a rolling window.
3. State Evolution: time-varying state coefficients, subtracted of the initial state vector.
4. Normal Q-Q Plot: we check if the observation follows the Gaussian distribution of estimated
mean and variance. To that end, we display a Q-Q plot of the residual divided by the estimated
standard deviation, against the standard normal distribution.
# S3 method for statespace
plot(x, pause = FALSE, window_size = 7, date = NULL, sel = NULL, ...)No return value, called to display plots.
the statespace object.
(default FALSE) if set to FALSE then the plots are displayed on a single
page, otherwise a new page is created for each plot.
(default 7) the window size of the rolling mean computed on the
error to display the bias, and on the mean squared error to display a rolling RMSE.
(default NULL) defines the values for the x-axis.
(default NULL) defines a subset of the data on which we zoom.
For instance one can display the evolution of the SSM on a test set and not the whole data set.
additional parameters