checkResiduals {dse} | R Documentation |
Calculate autocorrelation diagnostics of a time series matrix or TSdata or residuals of a TSestModel
checkResiduals(obj, ...)
## Default S3 method:
checkResiduals(obj, ac=TRUE, pac=TRUE, select=seq(nseries(obj)),
drop=NULL, plot.=TRUE, graphs.per.page=5, verbose=FALSE, ...)
## S3 method for class 'TSdata'
checkResiduals(obj, ...)
## S3 method for class 'TSestModel'
checkResiduals(obj, ...)
obj |
An TSestModel or TSdata object. |
ac |
If TRUE the auto-correlation function is plotted. |
pac |
If TRUE the partial auto-correlation function is plotted. |
select |
Is used to indicate a subset of the residual series. By default all residuals are used. |
drop |
Is used to indicate a subset of the residual time periods to drop. All residuals are used with the default (NULL).Typically this can be used to get rid of bad initial conditions (eg. drop=seq(10) ) or outliers. |
plot. |
If FALSE then plots are not produced. |
graphs.per.page |
Integer indicating number of graphs to place on a page. |
verbose |
If TRUE then the auto-correlations and partial auto-correlations are printed if they are calculated. |
... |
arguments passed to other methods. |
This is a generic function. The default method works for a time series matrix which is treated as if it were a matrix of residuals. However, in a Box-Jenkins type of analysis the matrix may be data which is being evaluated to determine a model. The method for a TSestModel evaluates the residuals calculated by subtracting the output data from the model predictions.
A list with residual diagnostic information: residuals, mean, cov, acf= autocorrelations, pacf= partial autocorrelations.
Diagnostic information is printed and plotted if a device is available. Output graphics can be paused between pages by setting par(ask=TRUE).
data("eg1.DSE.data.diff", package="dse")
model <- estVARXls(eg1.DSE.data.diff)
checkResiduals(model)