Last active
August 29, 2015 14:12
-
-
Save ky0on/e6e903a8af0b43b14d23 to your computer and use it in GitHub Desktop.
機構セミナー@2015年1月15日 に載せるグラフを生成するRスクリプト
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
## | |
## Compare methods for nonhierarchical clustering | |
## | |
## Please install following libraries in advance | |
## * LICORS | |
## * apcluster | |
## | |
## @athor kyon | |
## @created Jan. 2 2015 | |
## | |
# | |
# generate 4-calsses dataset | |
# | |
gen.dataset <- function(sd) { | |
n <- 25 # the number of data for each label | |
p1 <- cbind(rnorm(n=n, mean=1, sd=sd),rnorm(n=n, mean=1, sd=sd)) | |
p2 <- cbind(rnorm(n=n, mean=10, sd=sd),rnorm(n=n, mean=1, sd=sd)) | |
p3 <- cbind(rnorm(n=n, mean=1, sd=sd),rnorm(n=n, mean=10, sd=sd)) | |
p4 <- cbind(rnorm(n=n, mean=10, sd=sd),rnorm(n=n, mean=10, sd=sd)) | |
label <- as.factor(c(rep(1,n), rep(2,n), rep(3,n), rep(4,n))) | |
d <- as.data.frame(cbind(rbind(p1, p2, p3, p4), label)) | |
colnames(d) <- c('x', 'y', 'label') | |
return(d) | |
} | |
# | |
# run clustering and plot result | |
# | |
compare.plot <- function(d, label) { | |
#init | |
par(mfrow=c(2, 3)) | |
n_cluster <- length(levels(factor(label))) | |
#answer | |
plot(d, pch=as.integer(label), col=as.integer(label), main='Answer') | |
#k-means | |
d.km <- kmeans(d, n_cluster) | |
plot(d, pch=d.km$cluster, col=d.km$cluster, main='k-means') | |
#k-means++ | |
library(LICORS) | |
d.kmpp <- kmeanspp(d, n_cluster) | |
plot(d, pch=d.kmpp$cluster, col=d.kmpp$cluster, main='k-means++') | |
#x-means | |
source('http://www.rd.dnc.ac.jp/~tunenori/src/xmeans.prog') | |
d.xm <- xmeans(d) | |
plot(d, pch=d.xm$cluster, col=d.xm$cluster, main='x-means') | |
#Affinity Propagation | |
require('apcluster') | |
d.s <- negDistMat(d, r=2) | |
d.ap <- apcluster(d.s) | |
plot(d, pch=as.integer(factor(labels(d.ap))), col=as.integer(factor(labels(d.ap))), main='Affinity Propagation') | |
} | |
#init pdf | |
pdf('clustering.pdf') | |
#artificial dataset | |
for (sd in seq(1, 3, 0.5)) { | |
d <- gen.dataset(sd=sd) | |
compare.plot(d[,1:2], d$label) | |
} | |
#iris dataset | |
## compare.plot(iris[,1:2], iris$Species) | |
compare.plot(iris[,2:3], iris$Species) | |
#finalize | |
dev.off() |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment