predict.SemiParBIV: Prediction function

In GJRM: Generalised Joint Regression Modelling

Prediction function

Description

It takes a fitted gjrm object and, for each equation, produces predictions for a new set of values of the model covariates or the original values used for the model fit. Standard errors of predictions can be produced and are based on the posterior distribution of the model coefficients. This function is a wrapper for predict.gam() in mgcv. Please see the documentation of predict.gam() for full details.

Usage

## S3 method for class 'SemiParBIV'
predict(object, eq, ...)

Arguments

object	A fitted gjrm object.
eq	The equation to be considered for prediction.
...	Other arguments as in predict.gam() in mgcv.

WARNINGS

When type = "response" this function will provide prediction assuming that the identity link function is adopted. This means that type = "link" and type = "response" will produce the same results, which for some distributions is fine. This is because, for internal reasons, the model object used always assumes an identity link. There are other functions in the package which will produce predictions for the response correctly and we are currently working on extending them to all models in the package. For all the other type values the function will produce the correct results.

When predicting based on a new data set, this function can not return correct predictions for models based on a copula value of "C0C90", "C0C270", "C180C90", "C180C270", "G0G90", "G0G270", "G180G90", "G180G270", "J0J90", "J0J270", "J180J90" or "J180J270".

Author(s)

Maintainer: Giampiero Marra [e-mail]

See Also

gjrm

GJRM documentation built on July 9, 2023, 7:15 p.m.