Kaplan-Meier Estimator

Calls survival::survfit() during training to estimate the survival function at the unique event times from the training set. During prediction, each test observation is assigned the same survival probabilities at these times. The risk score (crank) is calculated as the sum of the cumulative hazard function (also called expected mortality) at the event times.

Dictionary

This Learner can be instantiated via the dictionary mlr_learners or with the associated sugar function lrn():

mlr_learners$get("surv.kaplan")
lrn("surv.kaplan")

Meta Information

• Task type: “surv”

• Predict Types: “crank”, “distr”

• Feature Types: “logical”, “integer”, “numeric”, “character”, “factor”, “ordered”

• Required Packages: mlr3, mlr3survival, survival

Parameters

Empty ParamSet

References

Kaplan EL, Meier P (1958). “Nonparametric Estimation from Incomplete Observations.” Journal of the American Statistical Association, 53(282), 457–481. doi:10.1080/01621459.1958.10501452 .

See also

Other survival learners: mlr_learners_surv.coxph

Super classes

mlr3::Learner -> mlr3survival::LearnerSurv -> LearnerSurvKaplan

Active bindings

native_model

(survival::survfit)
The fitted model.

Methods

Public methods

• LearnerSurvKaplan$new()

• LearnerSurvKaplan$importance()

• LearnerSurvKaplan$selected_features()

• LearnerSurvKaplan$clone()

Inherited methods

• mlr3::Learner$base_learner()
• mlr3::Learner$configure()
• mlr3::Learner$encapsulate()
• mlr3::Learner$format()
• mlr3::Learner$help()
• mlr3::Learner$predict()
• mlr3::Learner$predict_newdata()
• mlr3::Learner$print()
• mlr3::Learner$reset()
• mlr3::Learner$train()

---

Method new()

Creates a new instance of this R6 class.

Usage

LearnerSurvKaplan$new()

---

Method importance()

All features have a score of 0 for this learner. This method exists solely for compatibility with the mlr3 ecosystem, as this learner is used as a fallback for other survival learners that require an importance() method.

Usage

LearnerSurvKaplan$importance()

Returns

Named numeric().

---

Method selected_features()

Selected features are always the empty set for this learner. This method is implemented only for compatibility with the mlr3 API, as this learner does not perform feature selection.

Usage

LearnerSurvKaplan$selected_features()

Returns

character(0).

---

Method clone()

The objects of this class are cloneable with this method.

Usage

LearnerSurvKaplan$clone(deep = FALSE)

Arguments

deep

Whether to make a deep clone.

Examples

# Define the Learner
learner = lrn("surv.kaplan")
learner
#> 
#> ── <LearnerSurvKaplan> (surv.kaplan): Kaplan-Meier Estimator ───────────────────
#> • Model: -
#> • Parameters: list()
#> • Packages: mlr3, mlr3survival, and survival
#> • Predict Types: [crank] and distr
#> • Feature Types: logical, integer, numeric, character, factor, and ordered
#> • Encapsulation: none (fallback: -)
#> • Properties: importance, missings, selected_features, and weights
#> • Other settings: use_weights = 'use'

# Define a Task
task = tsk("lung")

# Stratification based on event
task$set_col_roles(cols = "status", add_to = "stratum")

# Create train and test set
part = partition(task)

# Train the learner on the train set
learner$train(task, row_ids = part$train)
learner$native_model
#> Call: survfit(formula = task$formula(1), data = task$data(cols = task$target_names), 
#>     weights = NULL)
#> 
#>        n events median 0.95LCL 0.95UCL
#> [1,] 112     81    310     267     390

# Make predictions for the test set
predictions = learner$predict(task, row_ids = part$test)
predictions
#> 
#> ── <PredictionSurv> for 56 observations: ───────────────────────────────────────
#>  row_ids time status    crank     distr
#>        9  567   TRUE 52.63894 <list[1]>
#>       13  301   TRUE 52.63894 <list[1]>
#>       21  473   TRUE 52.63894 <list[1]>
#>      ---  ---    ---      ---       ---
#>      158  185  FALSE 52.63894 <list[1]>
#>      159  222  FALSE 52.63894 <list[1]>
#>      168  177  FALSE 52.63894 <list[1]>

# Score the predictions
predictions$score()
#> surv.cindex 
#>         0.5