Accelerated Oblique Random Survival Forest Learner
mlr_learners_surv.aorsf.RdAccelerated oblique random survival forest.
Calls aorsf::orsf() from aorsf.
Note that although the learner has the property "missing" and it can in
principle deal with missing values, the behaviour has to be configured using
the parameter na_action.
Initial parameter values
n_thread: This parameter is initialized to 1 (default is 0) to avoid conflicts with the mlr3 parallelization.
mtry:This hyperparameter can alternatively be set via the added hyperparameter
mtry_ratioasmtry = max(ceiling(mtry_ratio * n_features), 1). Note thatmtryandmtry_ratioare mutually exclusive.
Prediction types
This learner returns three prediction types:
distr: a survival matrix in two dimensions, where observations are represented in rows and (unique event) time points in columns. Calculated using the internalpredict.ObliqueForest()function.response: the restricted mean survival time of each test observation, derived from the survival matrix prediction (distr).crank: the expected mortality usingmlr3proba::.surv_return().
Meta Information
Task type: “surv”
Predict Types: “crank”, “distr”, “response”
Feature Types: “integer”, “numeric”, “factor”, “ordered”
Required Packages: mlr3, mlr3proba, mlr3extralearners, aorsf, pracma
Parameters
| Id | Type | Default | Levels | Range |
| n_tree | integer | 500 | \([1, \infty)\) | |
| n_split | integer | 5 | \([1, \infty)\) | |
| n_retry | integer | 3 | \([0, \infty)\) | |
| n_thread | integer | 0 | \([0, \infty)\) | |
| pred_aggregate | logical | TRUE | TRUE, FALSE | - |
| pred_simplify | logical | FALSE | TRUE, FALSE | - |
| oobag | logical | FALSE | TRUE, FALSE | - |
| mtry | integer | NULL | \([1, \infty)\) | |
| mtry_ratio | numeric | - | \([0, 1]\) | |
| sample_with_replacement | logical | TRUE | TRUE, FALSE | - |
| sample_fraction | numeric | 0.632 | \([0, 1]\) | |
| control_type | character | fast | fast, cph, net | - |
| split_rule | character | logrank | logrank, cstat | - |
| control_fast_do_scale | logical | FALSE | TRUE, FALSE | - |
| control_fast_ties | character | efron | efron, breslow | - |
| control_cph_ties | character | efron | efron, breslow | - |
| control_cph_eps | numeric | 1e-09 | \([0, \infty)\) | |
| control_cph_iter_max | integer | 20 | \([1, \infty)\) | |
| control_net_alpha | numeric | 0.5 | \((-\infty, \infty)\) | |
| control_net_df_target | integer | NULL | \([1, \infty)\) | |
| leaf_min_events | integer | 1 | \([1, \infty)\) | |
| leaf_min_obs | integer | 5 | \([1, \infty)\) | |
| split_min_events | integer | 5 | \([1, \infty)\) | |
| split_min_obs | integer | 10 | \([1, \infty)\) | |
| split_min_stat | numeric | NULL | \([0, \infty)\) | |
| oobag_pred_type | character | risk | none, surv, risk, chf, mort | - |
| importance | character | anova | none, anova, negate, permute | - |
| importance_max_pvalue | numeric | 0.01 | \([1e-04, 0.9999]\) | |
| tree_seeds | integer | NULL | \([1, \infty)\) | |
| oobag_pred_horizon | numeric | NULL | \([0, \infty)\) | |
| oobag_eval_every | integer | NULL | \([1, \infty)\) | |
| oobag_fun | untyped | NULL | - | |
| attach_data | logical | TRUE | TRUE, FALSE | - |
| verbose_progress | logical | FALSE | TRUE, FALSE | - |
| na_action | character | fail | fail, omit, impute_meanmode | - |
References
Jaeger BC, Long DL, Long DM, Sims M, Szychowski JM, Min Y, Mcclure LA, Howard G, Simon N (2019). “Oblique random survival forests.” The Annals of Applied Statistics, 13(3). doi:10.1214/19-aoas1261 .
Jaeger BC, Welden S, Lenoir K, Speiser JL, Segar MW, Pandey A, Pajewski NM (2023). “Accelerated and interpretable oblique random survival forests.” Journal of Computational and Graphical Statistics, 1–16. doi:10.1080/10618600.2023.2231048 .
See also
as.data.table(mlr_learners)for a table of available Learners in the running session (depending on the loaded packages).Chapter in the mlr3book: https://mlr3book.mlr-org.com/basics.html#learners
mlr3learners for a selection of recommended learners.
mlr3cluster for unsupervised clustering learners.
mlr3pipelines to combine learners with pre- and postprocessing steps.
mlr3tuning for tuning of hyperparameters, mlr3tuningspaces for established default tuning spaces.
Super classes
mlr3::Learner -> mlr3proba::LearnerSurv -> LearnerSurvAorsf
Methods
Method oob_error()
OOB concordance error extracted from the model slot
eval_oobag$stat_values
Examples
# Define the Learner
learner = mlr3::lrn("surv.aorsf")
print(learner)
#> <LearnerSurvAorsf:surv.aorsf>: Oblique Random Forest
#> * Model: -
#> * Parameters: n_thread=1
#> * Packages: mlr3, mlr3proba, mlr3extralearners, aorsf, pracma
#> * Predict Types: [crank], distr, response
#> * Feature Types: integer, numeric, factor, ordered
#> * Properties: importance, missings, oob_error
# Define a Task
task = mlr3::tsk("grace")
# Create train and test set
ids = mlr3::partition(task)
# Train the learner on the training ids
learner$train(task, row_ids = ids$train)
print(learner$model)
#> ---------- Oblique random survival forest
#>
#> Linear combinations: Accelerated Cox regression
#> N observations: 670
#> N events: 217
#> N trees: 500
#> N predictors total: 6
#> N predictors per node: 3
#> Average leaves per tree: 40.414
#> Min observations in leaf: 5
#> Min events in leaf: 1
#> OOB stat value: 0.84
#> OOB stat type: Harrell's C-index
#> Variable importance: anova
#>
#> -----------------------------------------
print(learner$importance())
#> revascdays revasc los age stchange sysbp
#> 0.5842752 0.5224366 0.2233195 0.1837251 0.1193232 0.1039818
# Make predictions for the test rows
predictions = learner$predict(task, row_ids = ids$test)
# Score the predictions
predictions$score()
#> surv.cindex
#> 0.8578376