Question

How can I duplicate this style of graph, with multiple plots on one graph, and, preferably, legends attached as below.

I have tried the concept of “facet” but ggplot2 and trellis:xyplot both think of facets as separate panels rather than overlaid plots.

I can do it using plain Jane plot() and line().. but was using ggplot2 and woudl like to get multiple lines on one plot in that package.

Here is some example data in long form (captured from the plot using a nifty app called “Graphclick”)

comp <- read.table(pipe("pbpaste"), header=T, sep=',')

company, year, sales
Apple,1975.003,17298.457
Apple,1977.302,16784.502
Apple,1978.314,17298.457
Apple,1980.246,20730.098
Apple,1981.533,27608.426
Apple,1984.293,40862.852
Apple,1986.408,50468.617
Apple,1987.328,48236.188
Apple,1988.892,35676.547
Apple,1989.904,34616.582
Apple,1991.192,44732.742
Apple,1992.387,44732.742
Apple,1993.399,39055.324
Apple,1995.791,37894.922
Apple,1996.895,39648.746
Apple,1998.274,52804.367
Apple,1999.378,61399.512
Apple,2001.770,2.350e5
Apple,2005.265,7.735e5
Toshiba,1999.378,86856.6
Toshiba,2001.862,1.192e5
Toshiba,2004.069,1.495e5
Toshiba,2004.069,1.495e5
IBM,1975.003,22019.092
IBM,1975.830,27195.193
IBM,1976.934,30682.320
IBM,1978.130,31148.527
IBM,1980.430,35676.547
IBM,1981.625,35676.547
IBM,1983.005,39648.746
IBM,1985.305,40862.852
IBM,1986.408,46102.508
IBM,1987.512,64241.156
IBM,1989.996,75832.898
IBM,1991.100,84276.039
IBM,1992.295,85556.641
IBM,1993.307,79342.539
IBM,1994.779,79342.539
IBM,1995.791,84276.039
IBM,1996.895,95082.484
IBM,1996.895,95082.484
Commodore,1975.003,33588.051
Commodore,1975.830,34616.582
Commodore,1977.118,25219.982
Commodore,1978.130,23388.229
Commodore,1979.326,25992.234
Commodore,1980.521,21689.514
Commodore,1981.717,25219.982
Commodore,1984.201,6999.029
Commodore,1985.213,1670.460
Commodore,1986.408,1458.447

_{(source: asymco.com)}

Answer 1

so, in ggplot2, this code works

qplot(year, sales, data=comp, colour=as.factor(company), group= company, geom="path", log="y")

The only things left now is to format the value on the Y axis as numeric (not sci notation), and the labels are in an off-graph legend, rather than on the plots… Final suggestions welcomed.

This is a lot easier in the end than plot() + lines(), as that required support code to get the ranges, iterate over the group levels etc.