In ebridge2/badmf: Bayesian Decision-Making Forests (BaD-MFs)

require(badmf)
require(ggplot2)
require(MASS)
n=400
d=30

Bayesian Decision-Making Forest (BaD-MF)

In this notebook, we will investigate how to use the nearest centroid classifier.

We simulate 400 examples of 30 dimensional points:

# generate simulation data
testdat <- badmf.sims.rtrunk(n, d)
X <- testdat$X
Y <- testdat$Y

data <- data.frame(x1=X[,1], x2=X[,2], y=Y)
data$y <- factor(data$y)
ggplot(data, aes(x=x1, y=x2, color=y)) +
  geom_point() +
  xlab("x1") +
  ylab("x2") +
  ggtitle("Simulated Data") +
  xlim(-10, 10) +
  ylim(-10, 10)

Training

# reorder the points
set.seed(1234)
reord <- sample(n, replace=FALSE)
# split into a training and testing set
train.idx <- reord[1:(4*n/5)]; test.idx <- reord[(4*n/5+1):n]
X.train <- X[train.idx,,drop=FALSE]; Y.train <- Y[train.idx]
X.test <- X[test.idx,,drop=FALSE]; Y.test <- Y[test.idx]

set.seed(1234)
fit.rf <- rf.class.fit(X.train, Y.train)

set.seed(1234)
Y.pred <- predict(fit.rf, X.test)

print(sprintf("Misclassification Rate: %.3f", 1 - mean(Y.pred == Y.test)))

ebridge2/badmf documentation built on June 4, 2019, 8:53 a.m.