Plot the cluster member in r

问题

I use DTW package in R. and I finally finished hierarchical clustering. but I wanna plot time-series cluster separately like below picture.

sc <- read.table("D:/handling data/confirm.csv", header=T, sep="," )
rownames(sc) <- sc$STDR_YM_CD
sc$STDR_YM_CD <- NULL
col_n <- colnames(sc)

hc <- hclust(dist(sc), method="average")
plot(hc,  main="")

How can I do it?? My data in http://blogattach.naver.com/e772fb415a6c6ddafd1370417f96e494346a9725/20170207_141_blogfile/khm2963_1486442387926_THgZRt_csv/confirm.csv?type=attachment

回答1:

You can try this:

sc <- read.table("confirm.csv", header=T, sep="," )
rownames(sc) <- sc$STDR_YM_CD 
sc$STDR_YM_CD <- NULL
col_n <- colnames(sc)

sc <- t(sc) # make sure your rows represent the time series data
id <- rownames(sc)
head(sc)

hc <- hclust(dist(sc), method="average")
plot(hc,  main="")

n <- 20
sc <- cbind.data.frame(id=id, sc, cluster=cutree(hc, k = n))

library(dplyr)
library(tidyr)
library(ggplot2)
sc %>% gather(variable, value, -id, -cluster) %>%
ggplot(aes(variable, value, group=id, color=id)) + geom_line() + 
  facet_wrap(~cluster, scales = 'free') + guides(color=FALSE) + 
  theme(axis.text.x = element_text(angle=90, vjust = 0.5))

回答2:

You can use cutree to cluster the data points and use facet_wrap (from package ggplot2) on clusters to plot them. Since I couldn't get your data, I have an example from publicly available data.

narrest <- USArrests

# Clustering
hc <- hclust(dist(narrest), "ave")
plot(hc)

# Cut the tree to required number of clusters, here 3
narrest$clusters <- cutree(hc, k = 3)

# use facet_wrap from ggplot to one variable Murder 
d <- ggplot(narrest, aes(y=Murder, x=1:nrow(narrest))) + geom_line()
d + facet_wrap(~ clusters)
print(d)

来源：https://stackoverflow.com/questions/42081879/plot-the-cluster-member-in-r