nlar {tsDyn} | R Documentation |
Non-linear time series model, base class definition
Description
Generic non-linear autoregressive model class constructor.
Usage
nlar(
str,
coefficients,
fitted.values,
residuals,
k,
model,
model.specific = NULL,
...
)
Arguments
str |
a |
coefficients , fitted.values , residuals , k , model , model.specific |
internal structure |
... |
further model specific fields |
Details
Constructor for the generic nlar
model class. On a fitted object you
can call some generic methods. For a list of them, see
nlar-methods
.
An object of the nlar
class is a list of (at least) components:
- str
-
nlar.struct
object, encapsulating general infos such as time series length, embedding parameters, forecasting steps, model design matrix - coefficients
a named vector of model estimated/fixed coefficients
- k
total number of estimated coefficients
- fitted.values
model fitted values
- residuals
-
model residuals
- model
data frame containing the variables used
- model.specific
(optional) model specific additional infos
A nlar
object normally should also have a model-specific
subclass (i.e., nlar
is a virtual class).
Each subclass should define at least a print
and, hopefully, a
oneStep
method, which is used by predict.nlar
to
iteratively extend ahead the time series.
Value
An object of class nlar
. nlar-methods for a list of
available methods.
Author(s)
Antonio, Fabio Di Narzo
References
Non-linear time series models in empirical finance, Philip Hans Franses and Dick van Dijk, Cambridge: Cambridge University Press (2000).
Non-Linear Time Series: A Dynamical Systems Approach, Tong, H., Oxford: Oxford University Press (1990).
See Also
availableModels
for currently available built-in
models. nlar-methods for available nlar
methods.