predict.dbchoice {DCchoice} | R Documentation |
Predicting method for objects of class "dbchoice".
## S3 method for class 'dbchoice' predict(object, newdata = NULL, type = c("utility", "probability"), bid = NULL, ...)
object |
an object of class "dbchoice". |
newdata |
a data frame containing new data to predict. If |
type |
type of prediction (utility or probability). |
bid |
a bid design needed to predict with original data. |
... |
optional arguments. Currently not in use. |
The function predict()
for S3 object "dbchoice" calculates predicted values
according to the fitted model that is included in object
. The values are
predicted with the original data used for fitting the model if newdata = NULL
,
otherwise with a new data assigned to newdata
. There are two notes for
dbchoice()
: a bid design used for the fit must be assigned to bid
,
when predicting with the original data; the predicted values are calculated according
to the relevant single-bounded dichotomous choice model, when predicting with a new
data. See examples for details.
The current function does not estimate standard errors of predicted values.
When newdata = NULL
and type = utility
, a matrix containing utility
values under first (f
), second upper (u
), and second lower (l
) bids
is returned. When newdata = NULL
and type = probability
, a matrix
containing probabilities of choosing Yes-Yes (yy
), No-No (nn
), Yes-No (yn
),
and No-Yes (ny
) is returned. When a new data is assigned to newdata
,
predictions are calculated according to the relevant single-bounded dichotomous choice
model, and a vector containing utility values of choosing yes (type = utility
)
or probability of choosing yes (type = probability
) is returned.
## See Examples in dbchoice.