How to plot three point lines using ggplot2 instead of the default plot in R

Question

I have three matrix and I want to plot the graph using ggplot2. I have the data below.

library(cluster)
require(ggplot2)
require(scales)
require(reshape2)

data(ruspini)
x <- as.matrix(ruspini[-1])


w <- matrix(W[4,])

df <- melt(data.frame(max_Wmk, min_Wmk, w, my_time = 1:10), id.var = 'my_time')

ggplot(df, aes(colour = variable, x = my_time, y = value)) +
  geom_point(size = 3) +
  geom_line() +
  scale_y_continuous(labels = comma) +
  theme_minimal()

I want to add the three plots into one plot using a beautiful ggplot2. Moreover, I want to make the points with different values have different colors.

Answer 1

I'm not quite sure what you're after, here's a guess

Your data...

max <- c(175523.9, 33026.97, 21823.36, 12607.78, 9577.648, 9474.148, 4553.296, 3876.221, 2646.405, 2295.504)
min <- c(175523.9, 33026.97, 13098.45, 5246.146, 3251.847, 2282.869, 1695.64, 1204.969, 852.1595, 653.7845)
w <- c(175523.947, 33026.971, 21823.364, 5246.146, 3354.839, 2767.610, 2748.689,   1593.822, 1101.469, 1850.013)

Slight modification to your base plot code to make it work...

plot(1:10,max,type='b',xlab='Number',ylab='groups',col=3)
points(1:10,min,type='b', col=2)
points(1:10,w,type='b',col=1)

Is this what you meant?

If you want to reproduce this with ggplot2, you might do something like this...

# ggplot likes a long table, rather than a wide one, so reshape the data, and add the 'time' variable explicitly (ie. my_time = 1:10)
require(reshape2)
df <- melt(data.frame(max, min, w, my_time = 1:10), id.var = 'my_time')

# now plot, with some minor customisations...
require(ggplot2); require(scales)
ggplot(df, aes(colour = variable, x = my_time, y = value)) +
  geom_point(size = 3) +
  geom_line() +
  scale_y_continuous(labels = comma) +
  theme_minimal()

UPDATE after the question was edited and the example data changed, here's an edit to suit the new example data:

Here's your example data (there's scope for simplification and speed gains here, but that's another question):

library(cluster)
require(ggplot2)
require(scales)
require(reshape2)

data(ruspini)
x <- as.matrix(ruspini[-1])

wss <- NULL
W=matrix(data=NA,ncol=10,nrow=100)

for(j in 1:100){
  k=10  
  for(i in 1: k){
    wss[i]=kmeans(x,i)$tot.withinss    
  }
  W[j,]=as.matrix(wss)
}

max_Wmk <- matrix(data=NA, nrow=1,ncol=10)

for(i in 1:10){
  max_Wmk[,i]=max(W[,i],na.rm=TRUE)
}

min_Wmk <- matrix(data=NA, nrow=1,ncol=10)
for(i in 1:10){
  min_Wmk[,i]=min(W[,i],na.rm=TRUE)
}

w <- matrix(W[4,])

Here's what you need to do to make the three objects into vectors so you can make the data frame as expected:

max_Wmk <- as.numeric(max_Wmk)
min_Wmk <- as.numeric(min_Wmk)
w <- as.numeric(w)

Now reshape and plot as before...

df <- melt(data.frame(max_Wmk, min_Wmk, w, my_time = 1:10), id.var = 'my_time')

ggplot(df, aes(colour = variable, x = my_time, y = value)) +
  geom_point(size = 3) +
  geom_line() +
  scale_y_continuous(labels = comma) +
  theme_minimal()

And here's the result: