About
About SecondChance
SecondChance is a predictive analytics project built at UNC Charlotte to help animal shelter teams make better use of their data. We bring together public shelter records, clear visualizations, and predictive models to help staff understand adoption patterns, recognize animals that may need extra attention, and plan practical support.
01Why we chose this project
Why we chose this project
We chose this project because shelter data represents animals, staff decisions, and opportunities to help. We wanted to apply what we are learning in data science and analytics to a real problem: making useful patterns easier to see and turning those insights into earlier, more focused support for animals waiting for a home.
02Built by our team
Built by our team
Our team brings a shared background in data science and analytics through our work at UNC Charlotte. SecondChance brings together the areas we are studying: preparing data, building and evaluating predictive models, and communicating results in a way people can understand and use.
- Davis HigginsUNC Charlotte project team
- Bennett ShearinUNC Charlotte project team
- Seth SumterUNC Charlotte project team
- Hunter PaceUNC Charlotte project team
03Beyond Austin
Built with Austin data. Useful beyond Austin.
Austin gives us a public dataset for exploring how shelter records can support better planning. The same approach can help shelters across the United States understand waiting times, identify patterns, focus outreach, and evaluate changes to their practices. Shelter owners, staff, volunteers, rescue partners, and people interested in adoption can use the site to better understand the information behind shelter outcomes.
Each shelter serves a different community, so Austin's results are a starting point rather than a prediction for every shelter. A shelter-specific version would connect that organization's intake and outcome records, track its own trends, and train or check predictions against its local data. Staff would stay in control of decisions and use the results to guide additional support.
04How another shelter could use this approach
How another shelter could use this approach
A shelter-specific connection is a future adaptation. Today the site reads Austin's public feeds only; there is no sign-up and nothing to connect.
01
Connect its records
Intake and outcome events from the shelter's own system.
02
Validate and pair stays
One stay per arrival, closed by the outcome that names it; ambiguous pairs set aside.
03
Review local patterns
Waits, outcomes and pressure over time, with definitions attached.
04
Evaluate a local model
Train or recalibrate on local data; test on a later period; check by group.
05
Plan and measure outreach
Write the plan and its assumptions down, then compare what happened.
05What the project can and cannot do
What the project can and cannot do
SecondChance supports human judgment. It does not confirm adoption availability, diagnose an animal, or decide what care an animal should receive. Predictions are estimates, and a higher long-stay estimate should prompt more attention and support. This is a student research prototype and is not affiliated with the Austin Animal Center.
The source code is published under the MIT License. Austin's datasets are public domain. The Animaster motion components adapted for this site are licensed separately and are not redistributed; see the repository's ATTRIBUTION file.