Obtains marginal predictions/fitted measures for objects of class 'mvord'.
marginal_predict(
object,
newdata = NULL,
type = NULL,
subjectID = NULL,
newoffset = NULL,
...
)an object of class 'mvord'.
(optional) data frame of new covariates and new responses. The names of the variables should correspond to the names of the variables used to fit the model. By default the data on which the model was estimated is considered.
types "prob", "class", "linpred", "cum.prob", "all.prob" are available.
(optional) vector specifying for which subjectIDs the predictions
or fitted values should be computed.
(optional) list of length equal to the number of outcomes, each element containing a vector of offsets to be considered.
further arguments passed to or from other methods.
The following types can be chosen in marginal_predict:
type | description |
"prob" | predicted marginal probabilities for the observed response categories. Used as default if newdata contains column(s) for response variable(s). |
"class" | predicted marginal classes of the observed responses. |
"linpred" | predicted linear predictor |
"cum.prob" | predicted marginal cumulative probabilities for the observed response categories. |
"all.prob" | predicted marginal probabilities for all ordered classes of each response. Used as default if newdata contains no column(s) for response variable(s). |
If provided, the row names of the output correspond to the subjectIDs, otherwise they correspond to the row id of the observations.
predict.mvord, joint_probabilities