Tag label graph with geom_text in ggplot

Question

I created this graph with the following code:

library(ggplot2); library(reshape2); library(plyr) likert <- data.frame(age = c(rep("young", 5), rep("middle", 5), rep("old", 5)), score1 = c(rep("unlikely", 1), rep("likely", 1), rep("very likely", 13)), score2 = c(rep("disagree", 6), rep("neutral", 4), rep("agree", 5)), score3 = c(rep("no", 5), rep("maybe", 7), rep("yes", 3))) meltedLikert <- melt(dlply(likert, .(age), function(x) llply(x, table))) names(meltedLikert) <- c("score", "count", "variable", "age") ggplot(meltedLikert[meltedLikert$variable != "age",], aes(variable, count, fill=score)) + geom_bar(position="dodge", stat="identity") + geom_text(data=data.frame(meltedLikert), aes(variable, count, group=score, label=meltedLikert$score), size=4) + facet_grid(age ~ .) 

How can I tag position text so that each score label score above the corresponding row for the variable top of each row?

Answer 1

According to the answer in a related question, adding position = position_dodge(width=0.9) to the geom_text string calls the values:

 ggplot(meltedLikert[meltedLikert$variable != "age",], aes(variable, count, fill=score)) + geom_bar(position="dodge", stat="identity") + geom_text(data=data.frame(meltedLikert), aes(variable, count, group=score, label=meltedLikert$score), position = position_dodge(width=0.9), size=4) + facet_grid(age ~ .) 

However, I also wanted to point out a few other things. You should not use meltedLikert$score in aes() call; you should only refer to things in the data frame that are passed as data . In addition, meltedLikert already has data.frame , so calling data.frame() on it is not needed (although it does not hurt).

The real improvement is how you start creating your tabs. Instead, consider this:

 tabulatedLikert <- ldply(likert[-1], function(sc) { as.data.frame(table(age = likert$age, score = sc)) }) ggplot(tabulatedLikert, aes(x=.id, y=Freq, fill=score)) + geom_bar(position="dodge", stat="identity") + geom_text(aes(label=score), position=position_dodge(width=0.9), size=4) + facet_grid(age ~ .) 

You can correct the order of the bars by correcting them in the source data:

 likert2 <- mutate(likert, score1 = factor(score1, levels=c("unlikely", "likely", "very likely")), score2 = factor(score2, levels=c("disagree", "neutral", "agree")), score3 = factor(score3, levels=c("no", "maybe", "yes"))) tabulatedLikert2 <- ldply(likert2[-1], function(sc) { as.data.frame(table(age = likert2$age, score = sc)) }) ggplot(tabulatedLikert2, aes(x=.id, y=Freq, fill=score)) + geom_bar(position="dodge", stat="identity") + geom_text(aes(label=score), position=position_dodge(width=0.9), size=4) + facet_grid(age ~ .) 

Of course, at the moment, colors actually do not add anything, since everything is marked directly on the chart, so I’ll just get rid of them completely.

 ggplot(tabulatedLikert2, aes(x=.id, y=Freq, group=score)) + geom_bar(position="dodge", stat="identity", fill="gray70") + geom_text(aes(label=score), position=position_dodge(width=0.9), size=4) + facet_grid(age ~ .)