How to draw a plot joining points from two measure

问题:

First question here! I have two columns of data and each row are a pair of values. I want to plot the first column and the second column vertically and have a line connecting each pair of values, something that looks like this figure in the following link:

http://www.sciencedirect.com/science/article/pii/S0300957297000440#gr1

If you know how to do it, in any tools, such as R, or python, perl, excel, please let me know!

回答1:

And another R approach using matpoints and matlines (and boxplot)

dd <- data.frame(x=rnorm(15), y= rnorm(15))

boxplot(dd, boxwex = 0.3)
# note that you need to transpose `dd`
matpoints(y= t(dd), x= c(1.17,1.83),pch=19, col='black')
matlines(y= t(dd), x= c(1.2,1.8), lty=1, col = 'black')

回答2:

Here's an R approach with ggplot2, a bit quick and dirty:

library(ggplot2)

df <- data.frame(baseline=c(1,1,2,2,3,3,4,5,6,7,8,9,10,11),
                 sixmos  =c(5,6,5,7,8,9,10,12,12,2,1,5,2,3))

data <- data.frame(group = factor(1:nrow(df)), 
                   cat=c(rep('baseline',nrow(df)), 
                   rep('sixmos',nrow(df))), 
                   values=c(df$baseline,df$sixmos))

ggplot(data, aes(x=cat, y=values)) + 
  geom_line(aes(group=group)) + 
  geom_point(aes(group=group)) +
  geom_boxplot(data=df, aes(x='baselin', y=baseline)) + 
  geom_boxplot(data=df, aes(x='sixmos2', y=sixmos))

Also see this answer: Line charts by group

回答3:

Here is a very basic attempt in Python:

import pylab as pl

data = pl.array([[1,2],[2,3],[1,3],[2,1],[5,3],[3,2],[3,2],[1,1]])

first = data[:,0]
second = data[:,1]

xs = []
ys = []

for r in data:
   ys += list(r)
   ys.append(None)
   xs += [1.3,1.7]
   xs.append(None)

pl.plot([1.3]*len(first),first,'o',[1.7]*len(second),second,'o',xs,ys)
pl.boxplot(data)
pl.ylim([min(min(first),min(second))-.5,max(max(first),max(second))+.5])
labels = ("first", "second")
pl.xticks([1,2],labels)

pl.show()

will result in:

回答4:

Here is a proof of concept in R using segments. Cleaned up and added the boxplots in line with @mnel's answer:

first <- 1:10
second <- 2:11
boxplot(first,second, boxwex=0.3)
points(rep(c(1.2,1.8),each=10),c(first,second),pch=19)
segments(rep(1.2,10),first,rep(1.8,10),second,col="gray")

回答5:

Another option with R's lattice - not the tidiest one, but does the job:

#load packages
library(lattice)
library(latticeExtra)

#example data
B <- subset(OrchardSprays, treatment == "B")
D <- subset(OrchardSprays, treatment == "D")
BD <- rbind(B,D)

#create three separate plots
nobox = list(axis.line=list(col="transparent"))#to remove box around plots
boxplotB <- bwplot(decrease ~ treatment, B, ylab = NULL, ylim=c(0,70), 
                   par.settings=nobox)
boxplotD <- bwplot(decrease ~ treatment, D, ylab = NULL, ylim=c(0,70), 
                   par.settings=nobox)
plotBD <- xyplot(decrease ~ treatment, BD, col=1, ylim=c(0,70), pch=16,  
                 par.settings=nobox, panel=function(x, y, ...) {
                   panel.xyplot(x, y, ...)
                   panel.points(x, y, ...)
                   #this loop is required to create connections between points
                   for(i in 1:nrow(B)) 
                        panel.lines(1:2, c(y[i], y[i+nrow(B)]), alpha=0.5, ...)
                   }
                 )

#combine three plots
comb <- c(boxplotB, plotBD, boxplotD, layout = c(3,1), y.same = F)
update(comb, scales = list(at = list(NA, NA, NA), y = list(draw = FALSE)))