GTbasedIM (version 1.0.0)

IM_nodep: IM_nodep Function: Calculate Influence Measure for Features Without Dependency

Description

The IM_nodep function calculates the influence measure of equation (2) in Davila-Pena et al. (2024) when the partition set is P={{1},{2},{3},{4}}, which is equivalent to the influence measure in Datta et al. (2015).

Usage

IM_nodep(Xdata, Ydata)

Value

A vector of influences for each feature.

Arguments

Xdata

Matrix. A dataset where rows represent observations and columns represent features.

Ydata

Vector. The response variable associated with each row in Xdata.

Details

The IM_nodep function calculates the weighted average of the number of times a change in the value of a feature influences the response value.

References

Datta, A., Datta, A., Procaccia, A., & Zick, Y. (2015). Influence in classification via cooperative game theory. Proceedings of the Twenty–fourth International Joint Conference on Artificial Intelligence, 511–517. https://www.ijcai.org/Proceedings/15/Papers/078.pdf.

Davila-Pena, L., Saavedra-Nieves, A., & Casas-Méndez, B. (2024). On the influence of dependent features in classification problems: a game-theoretic perspective. arXiv preprint. tools:::Rd_expr_doi("10.48550/arXiv.2408.02481").

Examples

# Example usage from Example 5.2 in Davila-Pena et al. (2024):

library(CoopGame)
n.user <- 16
Xdata <- createBitMatrix(4)[,-5]
Xdata <- rbind(c(0,0,0,0),Xdata)
Ydata <- rep(0,n.user)
Ydata[1+c(10,11,13,14,15)] <- 1

# Scenario 1:
IM_nodep(Xdata,Ydata)