Suspicious Behavior at the Lekagul Preserve
May 2023
VAST 2017 Mini Challenge
The VAST challenge is an annual competition in the Visual Analytics community that accompanies the VIS conference. Each year, it consists of a series of related mini-challenges that participants must solve by creating original visual analysis systems. As a final class project, my partners and I tackled a mini-challenge from the 2017 iteration, in which mysterious behaviors are impacting the population of Rose Crested Pippets at the Lekagul Preserve.
The code for our system and our final presentation slides can be found here.
Topics
- Visual Analytics
- Clustering
- High-dimensional projection
Technologies Used
- D3
- Node
- React
- Scikit-learn
- PostgreSQL
Challenges & Learnings
This was my first time working with JavaScript frameworks and building a web-app from the ground up. In terms of our actual approach, one of my key contributions was to incorporate machine learning into the task. We transformed the data into vectors representing itineraries of different vehicles in the park and then used DBSCAN to cluster similiar itineraries and identify unusual ones. The other linked visuals in our system were all means of unpacking and exploring the contents of those itineraries.