Explain Joint Model Components via DALEX and survex

Extracts and creates explainers for all components of a joint_model: the survival model (via survex), the binary status model (via DALEX), and the continuous duration time model (via DALEX).

Usage

explain_joint(
  model,
  data = NULL,
  type = c("all", "survival", "status", "time", "stack"),
  times = NULL,
  label = NULL,
  verbose = FALSE,
  ...
)

explain_model(
  model,
  data = NULL,
  type = c("all", "survival", "status", "time", "stack"),
  times = NULL,
  label = NULL,
  verbose = FALSE,
  ...
)

explain_status(model, data = NULL, label = NULL, verbose = FALSE, ...)

explain_time(model, data = NULL, label = NULL, verbose = FALSE, ...)

explain_stack(model, data = NULL, label = NULL, verbose = FALSE, ...)

Arguments

model: A fitted joint_model object.
data: Evaluation data frame with the original predictor columns (before one-hot encoding; extra columns are ignored). If NULL, the training predictors are used, as the original columns. The status and time explainers re-apply the model's training encoding with forge, as predict.joint_model does, so a factor predictor is one column in the explainers too.
type: Character indicating which component(s) to explain: "all", "survival", "status", "time", or "stack".
times: Numeric vector of evaluation time points for survival explanations.
label: Optional custom label prefix.
verbose: Logical; if TRUE, progress messages are shown. Default is FALSE.
...: Additional arguments passed to the underlying explainers.

Value

If type = "all", a list of class "cbe_joint_explainer" containing $survival, $status, $time, and (if present) $stack explainers. Otherwise, the individual requested explainer.