Transform PredictionClassif to PredictionSurv by converting event probabilities of a pseudo status variable (discrete time hazards) to survival probabilities using the product rule (Tutz et al. 2016):

$$P_k = p_k\cdot ... \cdot p_1$$

Where:

• We assume that continuous time is divided into time intervals $$[0, t_1), [t_1, t_2), ..., [t_n, \infty)$$

• $$P_k = P(T > t_k)$$ is the survival probability at time $$t_k$$

• $$h_k$$ is the discrete-time hazard (classifier prediction), i.e. the conditional probability for an event in the $$k$$-interval.

• $$p_k = 1 - h_k = P(T \ge t_k | T \ge t_{k-1})$$

## Dictionary

This PipeOp can be instantiated via the dictionary mlr3pipelines::mlr_pipeops or with the associated sugar function mlr3pipelines::po():

PipeOpPredClassifSurvDiscTime$new() mlr_pipeops$get("trafopred_classifsurv_disctime")
po("trafopred_classifsurv_disctime")

## Input and Output Channels

The input is a PredictionClassif and a data.table with the transformed data both generated by PipeOpTaskSurvClassifDiscTime. The output is the input PredictionClassif transformed to a PredictionSurv. Only works during prediction phase.

## References

Tutz, Gerhard, Schmid, Matthias (2016). Modeling Discrete Time-to-Event Data, series Springer Series in Statistics. Springer International Publishing. ISBN 978-3-319-28156-8 978-3-319-28158-2, http://link.springer.com/10.1007/978-3-319-28158-2.

Other PipeOps: PipeOpPredTransformer, PipeOpTaskTransformer, PipeOpTransformer, mlr_pipeops_survavg, mlr_pipeops_trafopred_regrsurv, mlr_pipeops_trafopred_survregr, mlr_pipeops_trafotask_regrsurv, mlr_pipeops_trafotask_survclassif_disctime, mlr_pipeops_trafotask_survregr

Other Transformation PipeOps: mlr_pipeops_trafopred_regrsurv, mlr_pipeops_trafopred_survregr, mlr_pipeops_trafotask_regrsurv, mlr_pipeops_trafotask_survclassif_disctime, mlr_pipeops_trafotask_survregr

## Super class

mlr3pipelines::PipeOp -> PipeOpPredClassifSurvDiscTime

## Methods

### Public methods

Inherited methods

### Method new()

Creates a new instance of this R6 class.

#### Arguments

deep

Whether to make a deep clone.