Drew McNeely

As you may know, I put up the page at astria.tacc.utexas.edu/vr, which is a small demo of scatter plots using A-Frame. The plot shows the positions of objects tracked by AstriaGraph on October 30, with colors denoting clusters resulting from clustering analysis performed by DBSCAN. Since that page is shoddy at best, for the past couple of days I’ve been working to improve the vis. One thing I had to do was completely switch entity types in A-Frame: before I was using aframe-scatterplot, and I switched over to the aframe-point-component. [Read More]

For the past week, I’ve been busy cleaning up my project. Things tends to get rats-nest like, so I needed to take a break from generating results in order to stay organized. You may also have noticed that I got a new website! I’ll slowly migrate my hobbies and resume and stuff from the old site to this one. In organizing my code, I was finally able to develop a consistent method of writing the most populous clusters to files. [Read More]

Clustering Analysis of Space Objects using scikit-learn Yesterday and today, I spent time building up a database of space objects and performing clustering on them. Ultimately, I represented an orbit as (E, e, h) where E is specific energy, e is the Laplace-Runge-Lenz vector, and h is specific angular momentum. Pulling orbit data from AstriaGraph, I clustered these data points using the DBSCAN and OPTICS routines from Python’s scikit-learn library. [Read More]

Clustering Analysis on Space Objects Proper Representation of Space Objects As outlined in my grant proposal, a key step in comparing space objects is to find a representation that makes comparisons a natural endeavor irrespective of time or coordinate system. So far, I am simply representing each object as a point in trajectory space. Specifically, I have drawn a json tracklets dataset from AstriaGraph on 10/31/2019, and I am using each entry as a data point in my cluster analysis. [Read More]