1 library(ggplot2)
   2 library(plyr)
   3 
   4 ## get an interactive window
   5 quartz()
   6 ## x11()
   7 
   8 ## lexical decision data
   9 data(english, package="languageR")
  10 
  11 ## default scatterplot
  12 qplot(x = WrittenFrequency, y = RTlexdec, data = english,
  13       main = "Frequency and Reaction Time")
  14 
  15 ## transform the dependent variable
  16 qplot(x = WrittenFrequency, y = exp(RTlexdec), data = english)
  17 
  18 ## color for groups
  19 qplot(x = WrittenFrequency, y = RTlexdec, data = english,
  20       colour=AgeSubject)
  21 
  22 ## color for continuous variable
  23 qplot(WrittenFrequency, RTlexdec, data=english,
  24       colour=WrittenSpokenFrequencyRatio)
  25 
  26 ## shape for groups
  27 qplot(x = WrittenFrequency, y = RTlexdec, data = english,
  28       shape=AgeSubject)
  29 
  30 ## colour and shape for groups
  31 qplot(x = WrittenFrequency, y = RTlexdec, data = english,
  32       colour=AgeSubject,
  33       shape=WordCategory)
  34 
  35 ## colour, shape, and size
  36 qplot(x = WrittenFrequency, y = RTlexdec, data = english,
  37       colour=AgeSubject,
  38       shape=WordCategory,
  39       size=FamilySize)
  40 
  41 ## panels
  42 ## one conditioning variable
  43 qplot(x = WrittenFrequency, y = RTlexdec, data = english,
  44       colour=AgeSubject,
  45       shape=Frication,
  46       facets = . ~ WordCategory)
  47 
  48 ## panels with totals shown
  49 qplot(x = WrittenFrequency, y = RTlexdec, data = english,
  50       colour=AgeSubject,
  51       shape=AgeSubject,
  52       facets = . ~ WordCategory,
  53       margins=TRUE)
  54 
  55 ## two conditioning variables
  56 qplot(x = WrittenFrequency, y = RTlexdec, data = english,
  57       colour=AgeSubject,
  58       shape=Frication,
  59       facets = CV ~ WordCategory)
  60 
  61 ## panels with totals shown
  62 qplot(x = WrittenFrequency, y = RTlexdec, data = english,
  63       colour=AgeSubject,
  64       shape=Frication,
  65       facets = CV ~ WordCategory,
  66       margins=TRUE)
  67 
  68 ## changing type of plot requires changing the "geom"
  69 qplot(x = WrittenFrequency, data = english,
  70       geom = "density", fill=rev(AgeSubject))
  71 
  72 ## histograms don't have a sensible default for bin size
  73 qplot(x = RTlexdec, data = english,
  74       geom = "histogram",
  75       fill=AgeSubject)
  76 
  77 ## stacked bar charts based on counts are the default
  78 qplot(x=WordCategory, data = english,
  79       geom = "bar",
  80       fill=Voice)
  81 
  82 ## use "fill" positioning to convert them to proportions
  83 qplot(x=WordCategory, data = english,
  84       geom = "bar",
  85       fill=Voice,
  86       position="fill")
  87 
  88 ## use "dodge" positioning to unstack them
  89 qplot(x=WordCategory, data = english,
  90       geom = "bar",
  91       fill=Voice,
  92       position="dodge")
  93 
  94 ## use stat="identity" if you've already aggregated your data and want
  95 ## to plot the results
  96 d <- with(english, aggregate(RTlexdec, by=list(Age=AgeSubject, Category=WordCategory), FUN=mean))
  97 qplot(x=Age, y=x, data=d,
  98       fill=Category,
  99       geom="bar",
 100       stat="identity",
 101       position="dodge")
 102 
 103 ## plots from scratch, without qplot
 104 
 105 ## boxplot
 106 bxp <- ggplot(data = english, aes(x=AgeSubject, y=RTlexdec))
 107 bxp <- bxp + stat_boxplot(aes(fill=WordCategory))
 108 (bxp)
 109 
 110 ## barplot of means
 111 bpm <- ggplot(data=english, aes(x=AgeSubject, y=RTlexdec, fill=WordCategory))
 112 bpm <- bpm +
 113   stat_summary(fun.y=mean, geom="bar", pos="dodge")
 114 (bpm)
 115 
 116 ## change the scale on the y axis to zoom in on the relevant region of
 117 ## the data
 118 ## NB:  If you don't want to show the entire range starting 
 119 ## from 0, you should consider using a dot plot instead of a bar plot
 120 bpm <- bpm + scale_y_continuous(limits=c(5,7))
 121 (bpm)
 122 
 123 ## bar plot with automatically calculated normal error bars
 124 bpe <- ggplot(data=english, aes(x=AgeSubject, y=RTlexdec, fill=WordCategory))
 125 bpe <- bpe + stat_summary(fun.y="mean", geom="bar", pos="dodge")
 126 bpe <- bpe + stat_summary(fun.data="mean_cl_normal", geom="errorbar", width=0.2)
 127 (bpe)
 128 
 129 ## dot plot of means
 130 dpm <- ggplot(data=english, aes(x=AgeSubject, y=RTlexdec, colour=WordCategory))
 131 dpm <- dpm + stat_summary(fun.y="mean", geom="point")
 132 (dpm)
 133 
 134 ## dot plot of means with error bars of 2*sd
 135 dpe <- ggplot(data=english,
 136               aes(x=AgeSubject, y=RTlexdec, colour=WordCategory))
 137 dpe <- dpe + stat_summary(fun.data="mean_sdl", geom="pointrange")
 138 (dpe)
 139 
 140 ## horizontal dotplot, often used to display regression parameters
 141 dpe <- dpe + coord_flip()
 142 (dpe)
 143 
 144 ## xyplot with grouping by color
 145 xy.tr <- ggplot(data=english, aes(x=WrittenFrequency, y=RTlexdec, colour=AgeSubject))
 146 xy.tr <- xy.tr + geom_point()
 147 ## now add a specific color scale where we define an alpha level
 148 xy.tr <- xy.tr + scale_colour_hue(alpha=1/3)
 149 (xy.tr)
 150 
 151 ## add a smoother to our scatterplot
 152 xy.tr <- xy.tr + stat_smooth() 
 153 (xy.tr)
 154 
 155 ## or we can specify that the smooth be based on a linear model
 156 qplot(WrittenFrequency, RTlexdec, data=english, colour=AgeSubject) +
 157   scale_colour_hue(alpha=1/3) +
 158   stat_smooth(method=lm)
 159 
 160 ## and we can add a robust linear smooth if we want
 161 library(MASS)
 162 last_plot() + stat_smooth(method=rlm)
 163 
 164 ## apply the black and white theme to our plot
 165 last_plot() + theme_bw()
 166 
 167 ## and change font sizes.  See page 123 and 124 of the ggplot2 book.
 168 last_plot() + opts(title = "Frequency Effects",
 169                    plot.title=theme_text(size=24))
 170 
 171 
 172 ## finally, let's make a hexbinplot
 173 hbp <- ggplot(english, aes(x=WrittenFrequency, y=RTlexdec)) +
 174   stat_binhex() +
 175   stat_smooth(method=rlm, colour=I("orange"), size=1.5) +
 176   opts(panel.background=theme_blank())
 177 hbp