Support vector machines predicting either survival time, relative ranks or hybrid of risks. Calls survivalsvm::survivalsvm() from package survivalsvm.


R6::R6Class() inheriting from LearnerSurv.


Four possible models can be implemented, dependent on the model parameter. These correspond to predicting the prognosting index (PI) via regression, regression; predicting the PI by imposing a ranking constrant, vanbelle1, vanbelle2; a hybrid of the two hybrid. Whilst regression is the default, hybrid may be more efficient in tuning as it can be reduced to the previous depending on optimal parameters.




Van Belle, V., Pelcmans, K., Van Huffel S. and Suykens J. A.K. (2011a). Improved performance on high-dimensional survival data by application of Survival-SVM. Bioinformatics, 27, 87-94.

Van Belle, V., Pelcmans, K., Van Huffel S. and Suykens J. A.K. (2011b). Support vector methods for survival analysis: a comparison between ranking and regression approaches. Artificial Intelligence in Medicine, 53, 107-118.

See also


library(mlr3) task = tgen("simsurv")$generate(200) learner = lrn("surv.svm") resampling = rsmp("cv", folds = 3) resample(task, learner, resampling)
#> <ResampleResult> of 3 iterations #> * Task: simsurv #> * Learner: surv.svm #> * Performance: 0.480 [surv.harrells_c] #> * Warnings: 0 in 0 iterations #> * Errors: 0 in 0 iterations