PDFPlot

Package

GitHub: https://github.com/sauhaardac/pdfplot

PyPI: https://pypi.org/project/pdfplot/

Install instructions: pip install pdfplot

Background

When evaluating a machine learning model or analyzing a shifting dataset, you’ll often need to regenerate a set of figures multiple times. If you have many figures, this means saving them all into separate files, manually updating file names, etc. As a project grows, this can become pretty cumbersome.

That’s what my lightweight python package, PDFPlot, is for! It takes care of all of the boilerplate plotting so you can focus on making cool graphs instead of organizing them.

The best part is, you can create your plots anywhere in your code. As long as you call pdfplot.save_figs before the run is completed, all of your plots will be saved and organized.

Usage

import pdfplot

fig, ax = pdfplot.make_fig()
ax.plot([0, 1, 2, 3], [4, 3, 2, 1])
ax.set_title('First Plot plot!')

fig, ax = pdfplot.make_fig()
ax.plot([0, 1, 2, 3], [1, 2, 3, 4])
ax.set_title('Second plot!')

pdfplot.save_figs(folder='figures')  # magic

The save_figs method has a lot of built in magic. It does a few things by default:

• Automatically timestamps your figures with the date and time (sortable and human readable of course!)
• Creates a shortcut to the most recent file, named “latest.pdf”
• Opens the created pdf file in your default PDF viewer

These features can be toggled via arguments to the save_figs method.

You can also overload the make_fig method your plots to adjust all of your plots, simultaneously. If you want a grid, just update make_fig:

def make_fig():
    fig, ax = plt.subplots()
    fig.tight_layout()
    ax.grid(True)
    return fig, ax

pdfplot.make_fig = make_fig

Credits

Thanks to Benjamin Rivière for introducing me to the PDF generation functionality in Matplotlib!