Skip to contents

Shapley Explanations for SPQR / SPQRX Models using shapr

Source: R/SPQRX.R
eval.explain.shapr.Rd

Computes Shapley values for SPQR or SPQRX model predictions using the shapr package. Supports quantile (QF), cumulative distribution (CDF), and density (PDF) explanations.

Usage

eval.explain.shapr(
  model,
  x_training,
  x_explain,
  y_training,
  y_explain,
  type = "QF",
  tau = 0.5,
  shapley_method = "empirical",
  variable_names = NULL,
  original_output = FALSE
)

Arguments

model

A fitted model object returned by fit_spqrx().

x_training

Matrix or data frame of training covariates.

x_explain

Matrix or data frame of observations to explain.

y_training

Numeric vector of training responses.

y_explain

Numeric vector of response values used for CDF or PDF explanations.

type

Character string specifying explanation target: "QF", "CDF", or "PDF".

tau

Numeric quantile level in (0,1) used when type = "QF".

shapley_method

Character string specifying the shapr estimation approach (e.g., "empirical").

variable_names

Optional character vector of variable names. If NULL, default names are assigned.

original_output

Logical. If TRUE, returns the full shapr explanation object. If FALSE (default), returns a data frame of Shapley values.

Value

If original_output = FALSE, a data frame of Shapley values. Otherwise, the full shapr explanation object.