This week I was able to invert the data currently stored inside the persistence diagram by creating a new function that sorts everything such that the most important information is displayed in the longest bars (while less important features are now in the shorter bars). I’ve also drawn a second persistence diagram on the right side of the screen and changed the overall size of each barcode to fit both on the interface without taking up too much space. The PD on the right does not function correctly yet, but the goal is to store a second set of persistent homology features that the user can click on when interacting with the force-directed layout. Dr. Rosen and I hope that when selecting a feature on each side, we can view the graph in a new and interesting way.
I’ve also realized how difficult it is to gather useful data to analyze with our FDL. The current implementation of our FDL reads in and color codes the graph based on a group ID, but most of the data I’ve come across has no unique ID feature. However, Dr. Rosen believes this will not be a problem so I’m continuing looking for data regardless of a group ID feature. This week I also created a file I/O program to read in data and convert it to a JSON in a format that our code currently reads from.
Before we arrived at the current implementation, I tested a few different ways to display our data whenever a bar is selected in the persistence diagram. The picture featured on this post is when I tried aligning the vertices in a circle (ultimately not used!). We decided it would be useful to have edge-bundling for our graph, since the number of edges made it difficult to effectively get rid of the “hairball” we were trying to prevent with our layout.
Dr. Rosen and I have been looking at papers for edge bundling in java and force-directed layouts. We have something currently in place that is doing what we’d like, but there is a lot more work to do before I feel ready to submit our paper later this month. I also booked my flight and hotel for the IEEE Vis 2017 conference next month, which is exciting! I can’t wait to present our research and learn more about what people are doing with Visualization.
Thanks for reading!