tests/test02-logic.R

In Condens8R: Classes and Methods for Learning Complex Decision Tree Models

library(Condens8R)
##
nr <- 100 # features
nc <- 40  # samples
## check binary data predictors
set.seed(97531)
bimat <- matrix(rbinom(nr*nc, 1, 0.45), nrow = nr)
dimnames(bimat) <- list(paste0("B", 1:nr),
                        paste0("S", 1:nc))
LR <- c("L", "R")
stat <- factor(LR[1 + rbinom(nc, 1, 0.37)], levels = LR)

myMod <- learn(logicModeler, bimat, stat)
table(predict(myMod), stat) # mostly on the diagonal

rantest <- matrix(rbinom(nr*nc, 1, 0.45), nrow = nr)
dimnames(rantest) <- list(paste0("B", 1:nr),
                          paste0("S", 1:nc))
table(predict(myMod, rantest))

## Now check continous data that get dichotomized
comat <- matrix(rnorm(nr*nc, 0, 1), nrow = nr)
dimnames(comat) <- list(paste0("B", 1:nr),
                        paste0("S", 1:nc))
coMod <- learn(logicModeler, comat, stat)
table(predict(coMod), stat) # all on the diagonal
table(predict(coMod, rantest))