predict.tsglm: Predicts Method for Time Series of Counts Following...

In tscount: Analysis of Count Time Series

Predicts Method for Time Series of Counts Following Generalised Linear Models

Description

Predict future observations based on a fitted GLM-type model for time series of counts.

Usage

## S3 method for class 'tsglm'
predict(object, n.ahead=1, newobs=NULL, newxreg=NULL, level=0.95,
        global=FALSE, type=c("quantiles", "shortest", "onesided"),
        method=c("conddistr", "bootstrap"), B=1000,
        estim=c("ignore", "bootstrap", "normapprox", "given"), B_estim=B,
        coefs_given, ...)

Arguments

object	an object of class "tsglm". Usually the result of a call to tsglm.
n.ahead	positive integer value giving the number of steps ahead for which predictions should be made.
newobs	integer vector of known future observations of the time series. This argument is only relevant if more than one observation ahead is to be predicted (n.ahead greater than 1). The h-step-ahead prediction for h>1 is computed as a 1-step-ahead prediction given all previous values, which can be observations of the original time series or new observations provided in this argument. Previous observations which are not available are replaced by their respective 1-step-ahead prediction.
newxreg	matrix or vector containing new values for the covariates to be used for prediction. If newxreg is omitted or contains less rows than the value of n.ahead, the last known values of the covariates are used for prediction. This is usually not reasonable and it is strongly advised to explicitely make assumptions on future covariates and to specify the argument xreg accordingly.
level	numeric value determining the desired coverage rate of prediction intervals. If level=0 no prediction intervals are computed.
global	logical value saying whether the coverage rate for Y_{n+1},...,Y_{n+h} specified by argument level holds globally (global=TRUE) or for each of the n.ahead prediction intervals individually (global=FALSE, the default). In the former case the individual coverage rate for a single prediction interval is Bonferroni adjusted to a level of 1-(1-level)/n.ahead.
type	character value saying how the prediction interval shall be constructed. If type="quantiles" (the default), its limits are chosen to be the a- and (1-a)-quantiles of the respective (approximated) distribution, with a=(1-level)/2. If type="shortest" it is chosen such that it has minimal length. Note that these two types of construction principles frequently lead to the same result. If type="onesided" a one-sided prediction interval is constructed where the lower boundary is always zero.
method	character value saying which method to be used for computing the prediction intervals. If method="conddistr" the prediction intervals are based on the conditional distribution given by the model with the unknown parameters being replaced by their respective estimations. This is only possible if only 1-step-ahead predictions are to be computed (possibly recursively using the new observations given in argument newobs). If method="bootstrap" the predictive distribution is approximated by a parametric bootstrap where B trajectories of the process are simulated from the fitted model. This is currently only possible if no new observations are given in argument newobs. By default the method "conddistr" is prefered whenever it is appliccable.
B	positive integer value giving the number of samples of a parametric bootstrap to use for numerical determination of prediction intervals (only necessary if argument method="bootstrap").
estim	character value saying how the prediction intervals shall account for the additional uncertainty induced by the parameter estimation. This is particularly important if the model was fitted on a short time series. If estim="ignore" (the default), this additional uncertainty is ignored. The other two options (estim="bootstrap" and estim="normapprox") are only possible if method="bootstrap". If these are selected the bootstrap samples are not generated from a model with the parameters of the original fit. Instead, each of the B bootstrap samples is generated from a model with parameters which are itself randomly generated. This two-stage approach should take into account the additional estimation uncertainty. If estim="bootstrap", the parameters are obtained from a fit to a parametric bootstrap replication of the original time series. If estim="normapprox", the regression parameters are generated from a multivariate normal distribution which is based on the normal approximation of the original quasi maximum likelihood estimator and reflects the estimation uncertainty. In that case the additional distribution coefficients are not randomly generated such that their estimation uncertainty is ignored. If estim="given", the parameters are resampled from a table of possible parameters which need to be given in argument coefs_given.
B_estim	positive integer value giving the number of parameters used for resampling to account for estimation uncertainty. Only necessary for estim="bootstrap" and estim="normapprox". If B_estim is smaller than B, the parameters are resampled with replacement.
coefs_given	table with parameters in the rows. Only necessary for estim="given"). If nrow(coefs_given) is smaller than B, the parameters are resampled with replacement.
...	further arguments are currently ignored. Only for compatibility with generic function.

Details

Returns predictions for the n.ahead observations following the fitted time series contained in argument object. The 1-step-ahead prediction is the conditional expectation of the observation to be predicted given the past. The true parameters are replaced by their estimations given in argument object. For a 2-step-ahead-prediction the true previous observation is used when given in argument newobs, otherwise it is replaced by the 1-step-ahead prediction computed before. For a 3-step-prediction this holds for the previous two observations, which are replaced by their respective predictions if not available, and so on.

Unless level=0, the function also returns prediction intervals. Read the description of the arguments type andmethod for further details on the computation. Note that the prediction intervals do not reflect the additional uncertainty induced by the parameter estimation. However, for sufficiently long time series used for model fitting, it is expected that this uncertainty is negligible compared to the uncertainty of the predictive distribution. The argument estim allows to account fot this additional estimation uncertainty if method="bootstrap", see the description of this argument.

If prediction intervals are computed the function additionally returns the median of the predictive distribution. If method="conddistr" this is the analytical median of the conditional distribution, otherwise the empirical median of the simulated distribution.

Value

A list with at least the following element:

pred	a numeric vector of the predictions. Has class "ts" if the response used for fitting has this class.

If prediction intervals are calculated, the list has the additional element:

interval	a matrix with the columns "lower" and "upper" giving the lower and upper boundaries of prediction intervals for the future time points, each with an intended coverage rate as given in argument level. Has class "ts" if the response used for fitting has this class.
level	a numeric value determining the desired coverage rate of prediction intervals.
global	a logical value saying whether the coverage rate level holds globally or for each of the prediction intervals individually.
type	a character value saying how the prediction intervals were computed. Possible values are "quantiles" and "shortest".
method	a character value saying which method were used for computation of prediction intervals. Possible values are "conddistr" and "bootstrap".
B	an integer value giving the number of bootstrap samples which were used for computing prediction intervals. Is NULL if computation was done by method="conddistr".
estim	a character value saying how the prediction intervals account for estimation uncertainty of the model parameters. Possible values are "ignore", "bootstrap", "normapprox" and "given".
B_estim	an integer value giving the number of parameter values used for resampling to account for estimation uncertainty. This value is zero if the estimation uncertainty is ignored.
warning_messages	a character vector containing warning messages. This should be NULL if no warning messages occured.
median	a vector giving the median of the predictive distribution for each of the future time points. Has class "ts" if the response used for fitting has this class.
futureobs	a matrix (n.ahead, B) of bootstrap samples which were used for computing prediction intervals. Is NULL if computation was done by method="conddistr".

Author(s)

Tobias Liboschik and Philipp Probst

References

Liboschik, T., Fokianos, K. and Fried, R. (2017) tscount: An R package for analysis of count time series following generalized linear models. Journal of Statistical Software 82(5), 1–51, http://dx.doi.org/10.18637/jss.v082.i05.

See Also

tsglm for fitting a GLM for time series of counts.

Examples

###Campylobacter infections in Canada (see help("campy"))
campyfit <- tsglm(ts=campy, model=list(past_obs=1, past_mean=c(7,13)))
predict(campyfit, n.ahead=1) #prediction interval using conditional distribution
predict(campyfit, n.ahead=5, global=TRUE) #prediction intervals using parametric bootstrap

tscount documentation built on May 11, 2023, 3:04 p.m.