ParkPredict U — Monroe

a hackathon-built smart parking prototype that predicts the best campus lot based on time-of-day patterns and approach route, with AI-style recommendations and one-click Google Maps directions.

Comment

Inspiration## Inspiration

We all know the campus parking pain. You circle the same lots, waste 10–15 minutes, stress about being late, and still end up parking far. For a “smart” campus, it didn’t feel smart at all. So we thought: instead of guessing, what if we had a simple web app that tells you, right now, “go here, not there” for Monroe/ULM — no hardware, no paid APIs, just smart logic we can build in a few hours.

What it does

ParkPredict U looks at:

  • day of week
  • time you plan to arrive
  • which direction you’re coming from

Then it:

  • predicts how full each main lot is (A/B/C style lots),
  • highlights the best lot right now with a clean banner,
  • shows estimated free space + walk time,
  • and gives a one-click “Open in Google Maps” link straight to that lot.

It’s built as an AI-style, explainable predictor: it doesn’t just show colors, it tells you why that lot is recommended.

How we built it

  • Fully in the browser with HTML, CSS, and JavaScript.

  • Designed a lightweight “parking model” using simulated historical patterns:

    • core lots fill early,
    • overflow lots pick up later,
    • weekends and evenings calm down, etc.

  • Added a scoring layer that turns those patterns into a simple prediction for each lot.

  • Built a side panel “AI explanation” that generates a human-readable summary from the model.

  • Wired in Google Maps direction links (no paid API needed) so users can jump to navigation instantly.

  • Left a hook for the provided AI/ML key so it can call a real model later without changing the UI.

Challenges we ran into

  • Doing something that feels “real-time smart” without sensors, cameras, or a big budget.
  • Staying honest: we didn’t want to fake data, so we built a transparent model we can explain.
  • Keeping it hackathon-legal: all code started during the event, no secret backend, no expensive APIs.
  • Making the UI clean and fast enough that someone can understand it in 5 seconds during judging.

Accomplishments that we're proud of

  • A polished, deployed, zero-cost prototype that actually looks like a product, not just a demo.
  • Clear storytelling: “Here’s your best lot, here’s why, click here to go there.”
  • Campus-specific thinking for Monroe instead of a generic “AI parking” buzzword project.
  • Architecture that can plug in real data (gates, sensors, traffic) later with minimal changes.

What we learned

  • You can get a lot of “AI-feeling” usefulness just by combining good heuristics, UX, and explanations.
  • Constraints (no money, no hardware, short time) can actually push a cleaner design.
  • How important it is to show trust and transparency in smart city tools: people want to know why.

What's next for ParkPredict U

  • Hook into real Monroe/ULM data:

    • gate counters, camera-based occupancy, event schedules, weather.

  • Add live congestion inputs (when available) to refine predictions.

  • Expand to more lots + accessibility routes (priority for accessible parking & safe crossings).

  • Publish an API so other campus apps (shuttles, safety, events) can use the same predictions.

  • Turn this from “hackathon prototype” into a plug-and-play smart parking layer any campus can deploy.

What it does

How we built it

Challenges we ran into

Accomplishments that we're proud of

What we learned

What's next for ParkPredict U — Monroe

Built With

Share this project:

Updates