survivalmodels (version 0.1.9)

predict.akritas: Predict method for Akritas Estimator

Description

Predicted values from a fitted Akritas estimator.

Usage

# S3 method for akritas
predict(
  object,
  newdata,
  times = NULL,
  lambda = 0.5,
  type = c("survival", "risk", "all"),
  distr6 = FALSE,
  ...
)

Arguments

object

(akritas(1)) Object of class inheriting from "akritas".

newdata

(data.frame(1)) Testing data of data.frame like object, internally is coerced with stats::model.matrix(). If missing then training data from fitted object is used.

times

(numeric()) Times at which to evaluate the estimator. If NULL (default) then evaluated at all unique times in the training set.

lambda

(numeric(1)) Bandwidth parameter for uniform smoothing kernel in nearest neighbours estimation. The default value of 0.5 is arbitrary and should be chosen by the user.

type

(character(1)) Type of predicted value. Choices are survival probabilities over all time-points in training data ("survival") or a relative risk ranking ("risk"), which is the negative mean survival time so higher rank implies higher risk of event, or both ("all").

distr6

(logical(1)) If FALSE (default) and type is "survival" or "all" returns matrix of survival probabilities, otherwise returns a distr6::VectorDistribution().

...

ANY Currently ignored.

Value

A numeric if type = "risk", a distr6::VectorDistribution() (if distr6 = TRUE) and type = "survival"; a matrix if (distr6 = FALSE) and type = "survival" where entries are survival probabilities with rows of observations and columns are time-points; or a list combining above if type = "all".

Details

This implementation uses a fit/predict interface to allow estimation on unseen data after fitting on training data. This is achieved by fitting the empirical CDF on the training data and applying this to the new data.

References

Akritas, M. G. (1994). Nearest Neighbor Estimation of a Bivariate Distribution Under Random Censoring. Ann. Statist., 22(3), 1299<U+2013>1327. 10.1214/aos/1176325630

Examples

# NOT RUN {
if (requireNamespaces(c("distr6", "survival"))) {

library(survival)

train <- 1:10
test <- 11:20
fit <- akritas(Surv(time, status) ~ ., data = rats[train, ])
predict(fit, newdata = rats[test, ])

# when lambda = 1, identical to Kaplan-Meier
fit <- akritas(Surv(time, status) ~ ., data = rats[1:100, ])
predict_akritas <- predict(fit, newdata = rats[1:100, ], lambda = 1)[1, ]
predict_km <- survfit(Surv(time, status) ~ 1, data = rats[1:100, ])$surv
all(predict_akritas == predict_km)

# Use distr6 = TRUE to return a distribution
predict_distr <- predict(fit, newdata = rats[test, ], distr6 = TRUE)
predict_distr$survival(100)

# Return a relative risk ranking with type = "risk"
predict(fit, newdata = rats[test, ], type = "risk")

# Or survival probabilities and a rank
predict(fit, newdata = rats[test, ], type = "all", distr6 = TRUE)
}
# }