Is each row inside the area of applicability?
Examples
set.seed(1)
x_train <- matrix(rnorm(200), 100, 2) # 100 training rows of 2 scaled predictors
ref <- di_reference(x_train)
threshold <- aoa_threshold(ref, folds = rep(1:5, 20))
inside_aoa(dissimilarity_index(ref, rbind(c(0, 0), c(4, 4))), threshold)
#> [1] TRUE FALSE