We came to HackPrinceton to learn more about data visualization. In order to do this, we obtained NASA Meteorite Landing data and did a basic visualization using WebGL Globe. This forced us to clean our data, use feature scaling and log scaling, as well visualizing the size of each point based on mass.
Our data was taken from Kaggle under NASA Meteorite Landing with the following data:
- Classification data
The source contains over 45,000 different confirmed meteorite impacts over the time span of 300 A.D. to present day.
Our biggest challenge was understanding the development of multi-scene objects in WebGL, all while incorporating seamless render capabilities tied together with one WebGL perspective. One other challenge was properly scaling the data. Due to an extreme skew we decided to create a logarithmic scale. There are plenty of ways to do this, so the difficulty was determining which was most accurate for our demonstration purposes.
Our hope is to keep improving the project, having points show information after mouse-clicks, via hover-activity, and also utilizing massive data sources (more on AWS web-crawler Dec. 3rd). We also plan to render the simulation of this activity over the given time interval. During Rowan's MLH Hackday, we will be working on this project for a total of 12 hours. If time permits we'll be attempting to configure a complex web application. Something more long term would be to leverage real-time solar system data, such as data about the asteroid belt and Jupiter's location, against a predictive solution that provides simulation and associative figures regarding future phenomena.
We chose to split our time 50/50, not including sleep, so a lot of our time here was spent doing more than just competing. NOTE: it doesn't work on touch screens/mobile devices.