description
Box plots (as per John Tukey) represent five-number summary of sets of observations over different categories.
Boxes display lower/upper quartiles (Q1 and Q3) with the median line in the middle.
Lower (upper) whisker is defined by the highest (lowest) observation among those that are below Q1 - 1.5 * IQR (above Q3 + 1.5 * IQR) the boundary values included.
Finally, mild outliers are defined by the data points outside the whiskers but still within a radius of 3 * IQR. The rest of the outliers are defined as extreme.
Box plots are perfect for displaying summary statistics of data sets when the underlying distributions are unimodal (having a single mode). As such, they are good companions for simple hypotheses tests.
It should be noted that they hinder the details of the distributions. Therefore, if the observations have multimodal distributions a violin plot is a more appropriate choice.