Boxplot with Means

boxplots as alternative to barplot with error bars

• a more informative alternative to this barplot with errorbars
• using the movies example data set (part of the ggplot2 package)
• cut the year variable in 15 equal intervals gives the new variable years
• map this new year var to x, budget to y and colour also to years (aes(x=years,y=budget,colour=years))
• add a layer boxplot, set the fill colour for the bars to black
• format the y axis (use dollar formatting)
• change the x axis tick labels with a text theme: rotate the tick labels (angle=270), position adjustment with hjust and vjust
• we do not need a legend for he colours because the information is contained in the x axis, so get rid of it: opts(legend.position="none")

movies$years <- cut(movies$year,breaks=15)
ggplot(movies, aes(x=years,y=budget,colour=years)) +
  geom_boxplot(fill="black") +
  scale_y_continuous(labels=dollar) +
  stat_summary(fun.y="mean",geom="point") +
  opts(axis.text.x = theme_text(angle=270,hjust=0,vjust=1)) +
  opts(legend.position="none")

Warnmeldungen:
1: Removed 53573 rows containing non-finite values (stat_boxplot). 
2: Removed 53573 rows containing missing values (stat_summary).

Barplot with Errorbars

I think boxplots are the better alternative

• load the summarySE function
• and cut the year variable to create reasonable categories

library(ggplot2)
data(movies)
source("helpers.r")
movies$years <- cut(movies$year,breaks=15)
table(movies$years)

(1893,1900] (1900,1908] (1908,1915] (1915,1923] (1923,1930] (1930,1938] 
         65         162         276         327         935        3093 
(1938,1945] (1945,1953] (1953,1960] (1960,1968] (1968,1975] (1975,1983] 
       3725        3357        4116        3702        5023        4554 
(1983,1990] (1990,1998] (1998,2005] 
       6775        8257       14421

• extract the information

moviesSE <- summarySE(movies,measurevar="budget",groupvars="years",na.rm=T)

- a warning message because the one class without non-missing values in budget

Warnmeldung:
In qt(p, df, lower.tail, log.p) : NaNs wurden erzeugt

• a look at the new created data frame containing the means, se, sd and ci of the budget variable according to the years classes

head(moviesSE)

        years   N    budget        sd       se        ci
1 (1893,1900]   0       NaN        NA       NA        NA
2 (1900,1908]   2   1125.00   1590.99  1125.00  14294.48
3 (1908,1915]  11  44023.73  63415.40 19120.46  42603.05
4 (1915,1923]  25 314215.76 298129.12 59625.82 123061.65
5 (1923,1930]  76 703895.24 761287.07 87325.62 173961.55
6 (1930,1938] 205 588688.42 553872.30 38684.12  76271.96

• build the plot using this new data frame

p <- ggplot(moviesSE,aes(x=years,y=budget,colour=years)) 
p +
  geom_bar() +  ## add bars
  geom_errorbar(aes(ymin=budget-se,ymax=budget+se),width=0.5) + ## add errorbars
  opts(axis.text.x = theme_text(angle=315,hjust=0,vjust=1)) ## customize tick labels