A line is plotted for each possible configuration (3x3x3x3x2=162) Lines are colored and offset based on score. 

I use it to identify the best pipeline configuration in a ML experiment, based on an aggregated performance score.  

Haven't seen anything like this for python/matplot before and thought about putting it together as a package.

Any ideas on improvement? 

I would love to be able to visualize the variation across iterations. Any thoughts on how to achieve that? 

Posted by The–__–Dude

Share.

4 Comments

  1. Finally a diagram that shows how to squeeze spaghetti into a much too small pot without breaking them. Good work!

  2. Very nice. It would be also nice to be able to highlight a subset of lines (for example, color change the lines that pass through a subset of nodes) interactively to be able to visually pop out and explore elements of the graph, as it is pretty difficult to visually trace any particular line segment through the spaghetti.