# y-axis ticks of distribution plot

Can I know what are the ticks at y axis mean?

I created a distribution plot of titanic['Age'] data from.Kaggle Titanic Data How to learn more about distribution of age column from the dataset from the graph? Can anyone explain what are the ticks at y axis? How to learn from this plot.

• This is a case where a kernel density estimate adds nothing useful or interesting to a histogram. The Titanic passengers had whatever ages they had; there is, or was, not helpfully any wider population to infer to. Further, automated smoothing of the distribution so the zero bound is not respected is more then usually absurd here. (There are solutions for that.) Even if you take the line that the main problem is smoothing too much, the histogram is already informative and perfectly intelligible. Naturally, how to treat the Titanic data isn't the question. – Nick Cox Aug 10 '18 at 7:26
• This is a plot of a distribution; nevertheless (a) histogram and kernel density estimate are specific, definite terms (b) to some distribution plot might suggest a plot of a cumulative distribution. – Nick Cox Aug 10 '18 at 7:29

The scale on the y-axis is called "density".

A density is not a proportion (in particular it can be > 1).

Instead, this scale is such that the area under the histogram equals 1.

This scale is convenient for it allows the evaluation of probabilities.

For example the (estimated) probability of falling below 20 is the area of all rectangles on the left of 20.

From this plot, one can say, e.g., that the observed population density (here you can take "density" in its common use) is highest where the age is between 20 and 40 and lowest after 60.

• I ask students and colleagues to make this kind of calculation. Horizontal range of data: about 100 years. Rough mental average of vertical amounts: about 0.01. Result of guessing the area by thinking of a rectangle of equivalent area: 100 $\times$ 0.01 = 1. Inference about units: it has to be 100 years $\times$ 0.01 probability per year to yield 1 probability. Naturally I know the answer and the dialogue may include rougher approximation and/or more coaxing, but a reminder that the total area must be probability 1, as in this answer, is key. – Nick Cox Aug 10 '18 at 7:34