I've been getting some questions on how I created the sigmoid figure from my logistic regression post, in particular how I integrated LaTeX into matplotlib. It's quite easy so I thought I'd share some tips and break down how I made that plot.
You'll first need to make sure that you have an installation of LaTeX on your computer. Many of us are used to using Overleaf so if you haven't installed LaTeX locally I would go head and start here.
Outside of your typical matplotlib imports you'll need to enable an rcParam to enable LaTeX, this is highlighted in the below code block.
After that it's really simple to pass LaTeX style equations into your plotting kit. As an example, let's build up the prior figure.
Not that it's relevant to using LaTeX but the below code was used to compute points for plotting.
And for generating the plot itself we'll do go ahead and proceed in two steps.
To generate the primary plot I went ahead and used the following code. As you can see, to indicate you want to render a LaTeX equation all you need to do is pass a raw string with $$ around the equation.
This gives you the same plot you saw before without the written equation.
The additional equation box is nothing more than a matplotlib text box with some additional specifiers for styling.
Let's go through the arguments to understand what's going on here. Generally speaking you can find many more details on this here.
2, 0.18 are the position argumentsr$...$ is the input text, in this case an equationfontsize is self-explanatorybbox is the style specifier.boxstyle determines the bounding box styleec is a tuple of normalized RGB values for the border colorfc is the same as ec except for fill colorWith everything combined we get the original plot that is part of the original post.
If you have any questions feel free to reach out, I've enjoyed being able to help where I can. Let me know if you would be interested in a more in depth writeup regarding my usage of matplotlib and tips.