A hands-on AI engagement for the Helios Decision Center team: correlation, validation, and forecasting across 1,500+ Arizona K-8 schools, built on the tools they already use, with the team trained to run AI-driven analysis themselves.
At a glance
The challenge
The Helios Decision Center has rich public data on 1,500+ Arizona K-8 schools and runs its own analytics application. The team wanted to put statistical techniques to work on that data and to learn how to do it themselves with AI. Work like this had never been within reach for them before, so they also wanted proof that AI could now make it fast and affordable. Doing that combined the data science to surface correlations, outliers, and forecasts that are hard to do by hand, the data engineering to clean the data and validate it against official sources, and the software engineering to put it all into a working application, all of it driven by AI.
What we built
- A correlation and outlier analysis that compares any two of 60+ school metrics across every school in the dataset and flags the statistically unusual ones, in an interactive dashboard
- An evidence-first validation system that checks each field against its official public source (AZ School Report Cards, the Auditor General, the State Board of Education, the US Census) and logs every attempt for audit
- An exploratory predictive layer that forecasts per-school outcomes in math, ELA, and science, surfaces the main drivers behind each, and flags schools performing well above or below their prediction
- Hands-on training and reusable frameworks, taught on the team's own data and tools, so they can frame a problem for AI, check its output, and run these workflows themselves
- A portable, self-hostable stack that mirrors the application the team already uses, so the analyses fit how they work
The results
- Delivered all three analysis phases across the full 1,500+ school dataset: correlation and outlier detection, data validation, and prediction
- Showed the team what AI can do with their own data and stack, and trained them to design and run these workflows themselves
- Validation traces every field back to its official source and keeps an auditable record, so any number on the dashboard can be defended
- Forecasts come with the factors that drove them and read as decision support a person reviews, not a black box
- Moved fast and stayed repeatable, turning around each phase in days rather than the weeks or months this work used to take
- Delivered more value than the engagement cost, with every phase producing analysis the team could put to use right away
Built with
Kyle built an AI analysis framework across all 1,500+ Arizona K-8 schools, from correlation and outlier detection to evidence-first data validation and per-school forecasts, in a fraction of the time and cost it would have taken us otherwise. Just as valuable, my team and I came away able to design these AI workflows ourselves.