Prediction Object for Survival

This object stores the predictions returned by a learner of class LearnerSurv.

The task_type is set to "surv".

For accessing survival and hazard functions, as well as other complex methods from a PredictionSurv object, see public methods of survDistr and example below.

Super class

mlr3::Prediction -> PredictionSurv

Active bindings

truth

(Surv)
True (observed) outcome.

crank

(numeric())
Access the stored predicted continuous ranking.

distr

(survdistr::survDistr)
Returns the stored survival predictions as a survDistr distribution object. The survDistr API provides methods to transform survival data into other quantities such as cumulative hazard, hazard, probability density, and cumulative distribution functions.

lp

(numeric())
Access the stored predicted linear predictor.

response

(numeric())
Access the stored predicted survival time.

Methods

Public methods

• PredictionSurv$new()

• PredictionSurv$clone()

Inherited methods

• mlr3::Prediction$filter()
• mlr3::Prediction$format()
• mlr3::Prediction$help()
• mlr3::Prediction$obs_loss()
• mlr3::Prediction$print()
• mlr3::Prediction$score()

---

Method new()

Creates a new instance of this R6 class.

Usage

PredictionSurv$new(
  task = NULL,
  row_ids = task$row_ids,
  truth = task$truth(),
  crank = NULL,
  distr = NULL,
  lp = NULL,
  response = NULL,
  check = TRUE,
  data_type = "surv",
  interp_meth = "const_surv"
)

Arguments

task

(TaskSurv)
Task, used to extract defaults for row_ids and truth.

row_ids

(integer())
Row ids of the predicted observations, i.e. the row ids of the test set.

truth

(survival::Surv())
True (observed) response.

crank

(numeric())
Numeric vector of predicted continuous rankings (or relative risks). One element for each observation in the test set. For a pair of continuous ranks, a higher rank indicates that the observation is more likely to experience the event.

distr

(matrix())
A matrix of predicted survival probabilities, where rows = observations and columns correspond to time points.

lp

(numeric())
Numeric vector of linear predictor scores. One element for each observation in the test set. \(lp = X\beta\) where \(X\) is a matrix of covariates and \(\beta\) is a vector of estimated coefficients.

response

(numeric())
Numeric vector of predicted survival times. One element for each observation in the test set.

check

(logical(1))
If TRUE, performs argument checks and predict type conversions.

data_type

(character(1))
Type of data (e.g. survival or hazard) stored in distr.

interp_meth

(character(1))
Type of interpolation used in subsequent methods of the survDistr object.

---

Method clone()

The objects of this class are cloneable with this method.

Usage

PredictionSurv$clone(deep = FALSE)

Arguments

deep

Whether to make a deep clone.

Examples

library(mlr3)
task = tsk("lung")
learner = lrn("surv.kaplan")
part = partition(task)
p = learner$train(task, part$train)$predict(task, part$test)
head(as.data.table(p))
#>    row_ids  time status    crank     distr
#>      <int> <num> <lgcl>    <num>    <list>
#> 1:      15   371   TRUE 81.78041 <list[1]>
#> 2:      16   520   TRUE 81.78041 <list[1]>
#> 3:      18   118   TRUE 81.78041 <list[1]>
#> 4:      20    12   TRUE 81.78041 <list[1]>
#> 5:      29   460   TRUE 81.78041 <list[1]>
#> 6:      30   153   TRUE 81.78041 <list[1]>

p$distr # survDistr
#> A [55 x 107] survival matrix
#> Number of observations: 55
#> Number of time points: 107
#> Interpolation method: Piece-wise Constant Survival 

# survival probabilities of the 4 test observations at two time points
p$distr$survival(times = c(80, 100))[1:4, ]
#>             80       100
#> [1,] 0.9026549 0.8584071
#> [2,] 0.9026549 0.8584071
#> [3,] 0.9026549 0.8584071
#> [4,] 0.9026549 0.8584071