Inspiration

I was working on a project at an internship where I was running DNN models many different epochs within a Collaboratory notebook, recording the accuracy, and repeating this process until the accuracy plateaued. Then I would copy-paste it into a spreadsheet and use the built-in tool too create a scatterplot of the data. I realized I could instead create a cell in the notebook that created a matplotlib scatter plotting accuracy over iterations. I didn't like the default graph, and I wanted to be able to easily examine the other available styles. So I wrote this short Python module with 2 functions, one that lets you examine each style in order (displaying a scatter, bar, and line plot in one graph), and one that lets you type the style you want to examine.

What's next for matplotlibGraphTester

I could scale it up for more graph customization, like allowing users to use different colors or shapes.

Built With

Share this project:

Updates