dotplot | Telling a story through dots | Data Visualization library

This is generally easier to reproduce if you provide (a representative portion) of the actual data, e.g. the output of dput(head(TitanicSurvival, 100)). But as this is a fairly popular dataset, I had it already available.

The exact wording of your task may have led to some confusion here - geom_dotplot is essentially a variant of a histogram, and I believe the intention was to ask about plotting the datapoints (usually geom_point). Specifying the geom_pointrange function indicates that you should also show the range, but it's not actually asked for.

So, here is my take on it. First we prepare all statistics and labels we may want to plot. The trickiest step is the last one - setting the factor levels of the labels based on the order of the average age, so they will show up in that order on the y-axis:

I think you are overflowing the allocated space in the chart by using specified locations for each individual observation (sometimes called 'stacking'). Instead you should 'jitter' the positions of the individual observations inside a specific allocated region. Jittering, means to introduce a small amount of randomness to the position of a point to avoid (mostly anyhow) overplotting.

I will illustrate this using graphics from the core of R for the following fictitious data. This focuses attention on what is wrong, more than on the specific programming solution in ggplot, which I will let you work out.

The moral of the story is: you can't know the panel size exactly until the plot is being drawn. However, you can approximate it by subtracting the dimensions of plot decoration from the output dimension. This approach ignores that in reality, graphics devices can have a 'resolution'/'scaling' parameter that effects the true size of the text. The reasoning here is that since panels are 'null' units, they adapt to whatever is left of the output dimension, after every non-null units have been subtracted.

It actually is a stacking problem. The following solves:

The problem here is, that base plots directly draw on a device.

Please see this related answer for workarounds.

Your code is working fine once we switch to e.g. ggplot (only server.R needs to be modified):

This is a very interesting question. OP is looking to dodge geoms along the x axis, which is not typically difficult to do. The difficulty here lies in that you are dodging the same data using different geoms.

What you can do is use a bit of clever formatting, mapping, and faceting to recreate an example of the type of plot OP shows. For this example solution, I am using the built-in dataset, iris. In the future, OP, please be sure to provide a reproducible example using a built-in dataset, your data, or a sample of your data.

Here's the basic plot showing a dotplot on top of a box plot below - I'll be trying to split the boxplot on the right and dotplot on the left.

As others have mentioned, pivot_longer needs tidyr (i.e., tidyr::pivot_longer. Then, we can use . to pipe the my_data to the melt function (this ensures that var_1_col and var_2_col are dropped before going into melt). Then, we can specify the two new column names for the pivoted data (i.e., variable and value).

I want to modify [...], representing the values of two different variables.

If you're looking to plot two different variables on the same graph (and they share a common axis like the names in this case), you can construct two separate geom_point arguments.

If anyone comes across this, I figured it out with help from Dr. Revell, the creator of phytools. It turns out, having worked almost exclusively in ggplot2, I do not really know how to use png() properly!

This is how it should go: