clusterpro.varpro.RdClusterPro for unsupervised data visualization using Varpro rules.
clusterpro(data,
method = c("auto", "unsupv", "rnd"),
ntree = 100, nodesize = NULL,
max.rules.tree = 40, max.tree = 40,
papply = mclapply, verbose = FALSE, seed = NULL, ...)Data frame containing the usupervised data.
Type of forest used. Choices are "auto" (auto-encoder), "unsupv" (unsupervised analysis), and "rnd" (pure random forests).
Number of trees to grow.
Nodesize of trees. If not specified, value is set using an internal function optimized for sample size and dimension.
Maximum number of rules per tree.
Maximum number of trees used for extracting rules.
Use mclapply or lapply.
Print verbose output?
Seed for repeatability.
Further arguments to be passed to unsupv.
Unsupervised data visualization tool. A two-class analysis is run for a VarPro rule and its complementary region to obtain regression coefficients which identify variables of interest relative to the release variable. The regression coefficients are used to scale the centroids of the two regions. Save all such pairs of scaled centroids to obtain the enhanced learning data set for the release variable. Off the shelf data visualization tools can then be applied to the enhanced data to explore data relationships.
# \donttest{
##------------------------------------------------------------------
##
## V-cluster simulation
##
##------------------------------------------------------------------
vcsim <- function(m=500, p=9, std=.2) {
p <- max(p, 2)
n <- 2 * m
x <- runif(n, 0, 1)
y <- rep(NA, n)
y[1:m] <- x[1:m] + rnorm(m, sd = std)
y[(m+1):n] <- -x[(m+1):n] + rnorm(m, sd = std)
data.frame(x = x,
y = y,
z = matrix(runif(n * p, 0, 1), n))
}
dvc <- vcsim()
ovc <- clusterpro(dvc)
par(mfrow=c(3,3));plot(ovc,1:9)
par(mfrow=c(3,3));plot(ovc,1:9,col.names="x")
##------------------------------------------------------------------
##
## 4-cluster simulation
##
##------------------------------------------------------------------
if (library("MASS", logical.return=TRUE)) {
fourcsim <- function(n=500, sigma=2) {
cl1 <- mvrnorm(n,c(0,4),cbind(c(1,0),c(0,sigma)))
cl2 <- mvrnorm(n,c(4,0),cbind(c(1,0),c(0,sigma)))
cl3 <- mvrnorm(n,c(0,-4),cbind(c(1,0),c(0,sigma)))
cl4 <- mvrnorm(n,c(-4,0),cbind(c(1,0),c(0,sigma)))
dta <- data.frame(rbind(cl1,cl2,cl3,cl4))
colnames(dta) <- c("x","y")
data.frame(dta, noise=matrix(rnorm((n*4)*20),ncol=20))
}
d4c <- fourcsim()
o4c <- clusterpro(d4c)
par(mfrow=c(2,2));plot(o4c,1:4)
}
##------------------------------------------------------------------
##
## latent variable simulation
##
##------------------------------------------------------------------
lvsim <- function(n=1000, q=2, qnoise=15, noise=FALSE) {
w <- rnorm(n)
x <- rnorm(n)
y <- rnorm(n)
z <- rnorm(n)
ei <- matrix(rnorm(n * q * 4, sd = sqrt(.1)), ncol = q * 4)
e1 <- rnorm(n, sd = sqrt(.4))
e2 <- rnorm(n, sd = sqrt(.4))
wi <- w + ei[, 1:q]
xi <- x + ei[, (q+1):(2*q)]
yi <- y + ei[, (2*q+1):(3*q)]
zi <- z + ei[, (3*q+1):(4*q)]
h1 <- w + x + e1
h2 <- y + z + e2
dta <- data.frame(w=w,wi=wi,x=x,xi=xi,y=y,yi=yi,z=z,zi=zi,h1=h1,h2=h2)
if (noise) {
dta <- data.frame(dta, noise = matrix(rnorm(n * qnoise), ncol = qnoise))
}
dta
}
dlc <- lvsim()
olc <- clusterpro(dlc)
par(mfrow=c(4,4));plot(olc,col.names="w")
##------------------------------------------------------------------
##
## Glass mlbench data
##
##------------------------------------------------------------------
data(Glass, package = "mlbench")
dg <- Glass
## with class label
og <- clusterpro(dg)
par(mfrow=c(4,4));plot(og,1:16)
## without class label
dgU <- Glass; dgU$Type <- NULL
ogU <- clusterpro(dgU)
par(mfrow=c(3,3));plot(ogU,1:9)
# }