Boxplot in the default of R

Since no data was provided to resolve the problem, I will use the iris dataset to resolve it. This data set has 150 observations in 4 quantitative variables and 1 categorical one. This categorical variable has three levels. Because of these characteristics, this dataset becomes very interesting to illustrate how to solve this problem.

In addition, I will solve it in two ways: with the function boxplot , which is the default of R , and with the ggplot2 package.

First, we need to plot the simple boxplot using a quantitative variable as response and a categorical variable as a predictor:

boxplot(Petal.Length ~ Species, data=iris)

Next,itisnecessarytocalculatethemedianforeachspeciesofplant,sothattheboxplotsareconnectedaccordingtosomecriterion.Onewaytodothisiswiththeaggregatefunction:

aggregate(iris$Petal.Length,list(iris$Species),median)Group.1x1setosa1.502versicolor4.353virginica5.55

Tomakethechartquiteasier,I'llsavethisresultinsideanobjectcalledmedianas.Fromthis,simplyuselinesandplotthemedianscalculatedontopofthegeneratedboxplot,consideringthateachlevelofthevariableSpeciescanbeunderstoodasanumberfrom1to3:

medianas<-aggregate(iris$Petal.Length,list(iris$Species),median)lines(1:3,medianas$x)

Boxplotwithggplot2

Ithinkthissolutionismoreelegantbecausetheggplot2itselfcalculatesandplotsthestatisticswewanttoputonthegraphbyitself.Todothis,justusethestat_summaryfunction:

library(ggplot2)ggplot(iris,aes(x=Species,y=Petal.Length))+geom_boxplot()+stat_summary(fun.y=median,geom="line", lwd=1, aes(group=1))