Understanding Markersize in Python Scatter Plot

When creating scatter plots in Python using various libraries such as CairoMakie, one common issue that arises is understanding the `markersize` attribute. This attribute seems to scale the scatter size relative to the scatter marker's base size, but what does this mean exactly?

To visualize this concept, let's create a plot with different markers and see how they relate to each other. We'll use the CairoMakie library and create a scatter plot with various markers.

Firstly, it's essential to understand that not all markers are created equal. Some markers like Circle and Rect have their own backend implementations and can be faster to display than others. For example, special markers like 'A' or 'B' are not exactly markersize high or wide.

Here's an example plot showing different shapes accessible through the Symbols library as well as a few characters:

```python

import CairoMakie

sc = CairoMakie.scatter(1, 1, marker='A', markersize=50)

```

As you can see from this example, only special markers like Circle and Rect match the line width because their base size is 1 x 1. However, they don't quite match the Char or BezierPaths very well.

To better understand how `markersize` works, let's visualize some Chars, BezierPaths, Circle, and Rect in front of a line with a width of 50. This will help us see how these markers relate to each other.

When using CairoMakie, the default BezierPath markers like :circle, :rect, :utriangle, etc., have been chosen such that they approximately match Char markers of the same `markersize`.

The key takeaway from this is that `markersize` cannot be directly understood in terms of a unit like pixels (px), it depends on what is scaled. Therefore, one needs to consider the context and the plot being created when setting the marker size.

For instance, if you're using Char markers, `markersize` is equivalent to the font size when displaying the same characters using text. This means that increasing the `markersize` will make the characters larger, but it's not a direct mapping to pixels.

In conclusion, understanding `markersize` in Python scatter plots requires considering the context and the plot being created. By visualizing different markers and their relationships, you can better understand how to use this attribute effectively in your plots.

References:

* CairoMakie documentation

* Python Scatter Plot documentation

Markersize python scatter plot