library(randomForest)
## randomForest 4.6-10
## Type rfNews() to see new features/changes/bug fixes.
## Classification:
data(iris)
# ======================== MDS for 2D visualization
# calculate the distance matrix for first 4 columns
irisDist <- dist(iris[,1:4])
irisMds = cmdscale(irisDist, k=2)
# plot MDS result and color data points based on the 5th column
plot(irisMds)
text(irisMds[,1], (irisMds[,2]+max(irisMds[,2])*0.05), labels=1:nrow(irisMds), cex = 0.6, xpd = TRUE)
points(irisMds, col=c("red", "green", "blue")[as.numeric(iris$Species)])
# ========================= plot MDS and color data points based predication reslut (traing with origional dataset)
# random select 50 training samples
set.seed(1)
samp <- iris[sample(nrow(iris), 50), ]
iris.rf <- randomForest(Species ~ ., data=samp, proximity=TRUE, keep.forest=TRUE)
predicted<- predict(iris.rf, iris[,1:4])
plot(irisMds)
text(irisMds[,1], (irisMds[,2]+max(irisMds[,2])*0.05), labels=1:nrow(irisMds), cex = 0.6, xpd = TRUE)
points(irisMds, col=c("red", "green", "blue")[as.numeric(predicted)])
# ========================= plot MDS and color data points based predication reslut (traing on dataset after MDS)
irisMdsDf <- cbind(as.data.frame(irisMds), iris[,5])
names(irisMdsDf) <- c("x", "y", "Species")
samp <- irisMdsDf[sample(nrow(irisMdsDf), 50), ]
iris.rf <- randomForest(Species ~ ., data=samp, proximity=TRUE, keep.forest=TRUE)
predicted<- predict(iris.rf, irisMdsDf[,1:2])
plot(irisMds)
points(irisMdsDf[,1:2], col=c("red", "green", "blue")[as.numeric(predicted)])
# ========================= plot MDS and classification boundary
# create dataset with all possible data points
px <- seq(min(irisMdsDf$x), max(irisMdsDf$x), 0.02)
py <- seq(min(irisMdsDf$y), max(irisMdsDf$y), 0.02)
pgrid <-expand.grid (px, py)
names (pgrid) <- c ("x", "y")
predicted<- predict(iris.rf, pgrid)
z <- matrix(as.numeric(predicted), length(px), length(py))
plot(irisMdsDf[,1:2], col=c("red", "green", "blue")[as.numeric(irisMdsDf$Species)] )
contour (px, py, z , col="purple", drawlabels=F, add=T)
.
