spvim_ics: Influence function estimates for SPVIMs

In bdwilliamson/nova: Perform Inference on Algorithm-Agnostic Variable Importance

Influence function estimates for SPVIMs

Description

Compute the influence functions for the contribution from sampling observations and subsets.

Usage

spvim_ics(Z, z_counts, W, v, psi, G, c_n, ics, measure)

Arguments

Z	the matrix of presence/absence of each feature (columns) in each sampled subset (rows)
z_counts	the number of times each unique subset was sampled
W	the matrix of weights
v	the estimated predictiveness measures
psi	the estimated SPVIM values
G	the constraint matrix
c_n	the constraint values
ics	a list of influence function values for each predictiveness measure
measure	the type of measure (e.g., "r_squared" or "auc")

Details

The processes for sampling observations and sampling subsets are independent. Thus, we can compute the influence function separately for each sampling process. For further details, see the paper by Williamson and Feng (2020).

Value

a named list of length 2; contrib_v is the contribution from estimating V, while contrib_s is the contribution from sampling subsets.

bdwilliamson/nova documentation built on Feb. 1, 2024, 10:04 p.m.