inst/examples/xgb_iris.R

In viadee/anchorsOnR: AnchorsOnR: High-Precision Model-Agnostic Explanations in R

library(datasets)
library(xgboost)
data(iris)

x = as.matrix(iris[, 1:4])
y = as.numeric(factor(iris[, 5]))-1

param <- list(objective = "multi:softmax", num_class = 3)
model <- xgboost(params = param, data = x, label = y, nrounds = 10)
# Prepare bins:
# a list of bins per column, the nth-entry provides the bins of the nth-column
# Per bin: define whether numeric, the cuts
# of the bins, and whether the right or the left border of a bin is included
bins = list()
bins[[1]] = list(cuts = c(4.3, 5.4, 6.3, 7.9))
bins[[2]] = list(cuts = c(2.0, 2.9, 3.2, 4.4))
bins[[3]] = list(cuts = c(1, 2.633333, 4.9, 6.9))
bins[[4]] = list(cuts = c(0.1, 0.8666667, 1.6, 2.5))

# Prepare explainer to explain model with anchors
explainer = anchors(iris, model, target = "Species", bins = bins, tau = 1)

# Construct the explanations
explanations = explain(iris[sample(nrow(iris),3),], explainer)

# print explanations
printExplanations(explainer, explanations)

plotExplanations(explanations)