predict.CopulaCenR {CopulaCenR} | R Documentation |
Predictions from CopulaCenR regression models
Description
Predictions for new observations based on ic_spTran_copula
, rc_spCox_copula
,
ic_par_copula
and rc_par_copula
.
Usage
## S3 method for class 'CopulaCenR'
predict(object, newdata, type = "lp", ...)
Arguments
object |
a |
newdata |
a data frame (see details) |
type |
|
... |
further arguments |
Details
For the newdata
, when type = "survival"
, it must be a data frame with columns
id
(subject id), ind
(1,2 for two margins),
time
(to be evaluted) and covariates
;
when type = "lp"
, the newdata needs to have id
,
ind
and covariates
, but time
is not needed.
When the argument type = "lp"
, it gives a linear predictor for
each margin (i.e., log hazards ratio in the proportional hazards model,
log proportional odds in the proportional odds model).
When the argument type = "survival"
, the marginal and joint survival values
will be evaluated at the given time points in the newdata
.
Value
If type = "lp"
, it returns a data frame with id
,
lp1
(linear predictor for margin 1), lp2
.
If type = "survival"
, it returns a data frame with id
,
t1
(evaluated times for the margin 1), t2
,
S1
(predicted marginal survival probabilities for margin 1),
S2
and
S12
(the predicted joint survival probabilities at t1, t2
)
Examples
data(AREDS)
# fit a Copula2-Sieve model
copula2_sp <- ic_spTran_copula(data = AREDS, copula = "Copula2",
l = 0, u = 15, m = 3, r = 3,
var_list = c("ENROLLAGE","rs2284665","SevScaleBL"))
# Predicted probabilities for newdata
newdata = data.frame(id = rep(1:3, each=2), ind = rep(c(1,2),3),
time = c(2,3,5,6,7,8),
SevScaleBL = rep(3,6),
ENROLLAGE = rep(60,6),
rs2284665 = c(0,0,1,1,2,2))
output <- predict(object = copula2_sp, newdata = newdata)