ShapleyMats: Compute a matrix with SHAP values of each rule
In pre: Prediction Rule Ensembles

ShapleyMats

R Documentation

Compute a matrix with SHAP values of each rule

Description

This function computes the SHAP values of each rule and arranges them in a matrix that can be used to compute SHAP values of the model as a whole.

Usage

ShapleyMats(
  data,
  data_test = data,
  Rs,
  Rs_test = Rs,
  id_mat,
  interactions = FALSE
)

Arguments

`data`	a dataframe containing the data used to estimate the expectations in SHAP values
`data_test`	a dataframe containing the points to compute the SHAP values of. By default, this coincides with the datapoints used to estimate SHAP values.
`Rs`	a list of as many matrices as there are rules to compute the SHAP values for. The `j`-th element is a matrix corresponding to the `j`-th rule. Each of its columns corresponds to the 0-1 encoding of a subrule of the `j`-th rule, as observed in the data provided with input parameter `data`. Can be computed with the RuleMats function.
`Rs_test`	same as Rs, but computed for the (possibly different) observations provided from the `data_test` parameter.
`id_mat`	A matrix with as many rows as there are rules and as many columns as there are predictors. The `j`-th column has ones on the entries corresponding to rules that involve the `j`-th predictor, while all remaining entries are zeroes. Can be computed with the RuleMats function.
`interactions`	A logical parameter determining whether interaction SHAP values should also be computed.

Details

Code written by and used with permission from Giorgio Spadaccini.

Value

marginal A matrix with n\cdot p rows and as many columns as there are terms (both linear and rules). It is obtained by vertically stacking matrices of n rows. Each submatrix focuses on the SHAP values of a different predictor: the (i,k)-th entry of the j-th of such submatrices represents the contribution of the k-th term to the SHAP value of the i-th datapoint for the j-th predictor. The first p terms are the linear terms, and the remaining columns refer to the rules.

interaction A matrix with n\cdot p^2 rows and as many columns as there are terms (both linear and rules). It is obtained by vertically stacking matrices of n \cdot p rows. Each submatrix is in turn split into p subsubmatrices which focuses on the interaction SHAP values of a different pair of predictors: the (i,k)-th entry of the j-th subsubmatrix of the j'-th submatrix represents the contribution of the k-th term to the SHAP value of the i-th datapoint for the interaction between the j-th and the j'-th predictor. The first p terms are the linear terms, and the remaining columns refer to the rules.