如何绘制包含多个观察对象的时间序列表

Question

First, I'm a kind of a greenhorn in R. Unfortunately I couldn't find an existing tag which is related to my problem. I have several data.frames of the following structure:

Lines: City1, City2, City3, (...)

Columns: Migration-Flow 2004, Migration-Flow 2005, Migration-Flow 2006 (...)

City    Migration-Flow 2004    Migration-Flow 2005    Migration-Flow 2006
----    -------------------    -------------------    -------------------
City1   5920                   1339                   -3394
City2   3001                   4129                   4430
...

My lines (not my columns) representing a times series of each city. For instance, data.frame[,2] represents the time series of City1 between 2004 and 2010.

What I want to do is, to plot every line (Migration-Flow of each city from 2004 to 2010) in one single chart in order to compare them. Are there any ways to do this?

Answer 1

The brute-force non-elegant method (but why not, it's short):

# Generate artificial data
flow = matrix(runif(100,1000,2000),nrow=10)
d = data.frame(City=paste0("City",1:10),flow)

# Brute force
plot(unlist(d[1,-1]), type = "l")
for (i in 2:nrow(d)) lines(unlist(d[i,-1]),col=i)

Note that in your example data.frame[2,] is for City 1.

Answer 2

Using the plot command, you can generate a plot from the first row of your data. Then, using lines , you can add a new line to your plot for each new row. A simple for loop would make this process relatively simple. A trivial example is below:

set.seed(519310)
testdata <- data.frame(City=c("Trenton","New Dehli","Kabul","New York","Santa Fe","Berlin","Oslo"),
                       Migration.2004=sample(x = -1000:1000,size = 7,replace = TRUE),
                       Migration.2005=sample(x = -1000:1000,size = 7,replace = TRUE),
                       Migration.2006=sample(x = -1000:1000,size = 7,replace = TRUE),
                       Migration.2007=sample(x = -1000:1000,size = 7,replace = TRUE),
                       Migration.2008=sample(x = -1000:1000,size = 7,replace = TRUE),
                       Migration.2009=sample(x = -1000:1000,size = 7,replace = TRUE),
                       Migration.2010=sample(x = -1000:1000,size = 7,replace = TRUE))

plot(x = 1:(ncol(testdata)-1),
     y = testdata[1,2:8],
     type = "l",
     ylim = c(-1000,1000))

colors <- c("red","blue","green","yellow","purple","orange","dark blue")

for(x in 2:nrow(testdata)) {

    lines(x = 1:(ncol(testdata)-1),y = testdata[x,2:8],type = "l",col = colors[x])
}

This produces the output below:

Using the legend command you could produce a legend, and using colorRamp would let you have the colors dynamically change based on the number of rows. Arguments to xlab in plot would allow you to make the x-axis labels more complex. While this code is simple and relies on a specific number of rows, you could easily generalize it to work on more varied data.