Inspiration
Housing is often treated as the primary solution for recovery stability, reentry success, and long-term treatment outcomes. Yet across public data and lived experience, housing alone consistently plateaus.
I was inspired to build this project by a simple question:
Why does housing help—but not enough?
Working in recovery and reentry contexts, I’ve repeatedly seen individuals with housing still struggle, while others without strong formal services remain stable because of peer, family, faith, or community support. Public datasets rarely capture those dynamics.
This project explores that gap.
What it does
Recovery Stability Indicators (RSI) Analysis is an exploratory analytical framework that examines how recovery stability emerges from interactions between housing and support systems, rather than from isolated interventions.
Using a synthetic dataset modeled after public treatment datasets (e.g., TEDS-D, N-SSATS), the analysis:
- Tests whether housing and support act additively or multiplicatively
- Identifies where stability breaks down despite housing
- Surfaces system-level failure modes tied to treatment history and discharge context
- Explicitly documents what public datasets cannot measure
- Demonstrates how the framework can extend once missing variables become available
The goal is insight, not prediction.
How we built it
The project was built entirely in Hex, using Python for analysis and visualization.
Key steps included:
- Designing a synthetic dataset aligned with real public data structures
- Creating interpretable interaction analyses (Housing × Support)
- Visualizing treatment history effects on stability
- Analyzing treatment setting and discharge barriers
- Adding a conservative scenario analysis to demonstrate extensibility without overstating claims
All analysis is intentionally transparent and reproducible.
Challenges we ran into
The largest challenge was measurement limits, not tooling.
Public datasets systematically under-measure:
- Peer support
- Faith-based or community support
- Family support
- Justice-involved reentry navigation
Rather than ignoring this, the project treats the absence of these variables as a first-class analytical finding and designs around it.
Another challenge was balancing rigor with interpretability—ensuring non-technical judges could immediately understand what each result meant.
Accomplishments that we're proud of
- Demonstrating that housing × support is multiplicative, not additive
- Showing visually where housing fails to overcome accumulated treatment history
- Quantifying the cost of missing variables without pretending to observe them
- Keeping the analysis honest, readable, and policy-relevant
- Building a framework designed to grow with better data, not collapse without it
What we learned
We learned that many recovery analyses fail not because they are incorrect—but because they are incomplete.
When informal supports are invisible, systems misattribute failure to individuals instead of structure. Making those gaps visible changes how outcomes are interpreted and where interventions should focus.
We also learned how powerful clear visualization is in communicating complex system behavior.
What's next for Recovery Stability Indicators (RSI) Analysis
Next steps include:
- Applying the framework to real TEDS-D / N-SSATS data
- Extending the model to reentry-specific contexts
- Incorporating measured informal supports where available
- Using RSI as an evaluation layer for policy and program design
This project is intended as infrastructure for better questions, not a final answer
Log in or sign up for Devpost to join the conversation.