Project: Forest Refuge

Machine learning for cryptid real estate

Product Vision

"If Bigfoot exists, it's gotta live somewhere. Let's model the perfect hairy neighbor—before deforestation turns its home into a Starbucks parking lot."

Environmental Variables

Factor Data Source ML Weight Why It Matters
Forest Canopy Density NOAA 0.35 Bigfoot's not hanging out in parking lots (probably)
Water Source Proximity USGS 0.25 Even cryptids need to hydrate
Human Encroachment Census/OpenStreetMap 0.30 Inverse correlation with credible sightings
Prey Animal Density State Wildlife Dept APIs 0.10 Gotta eat something besides hikers

Game Plan

Data Pipeline

  • Ingest 10+ years of environmental data via PySpark
  • Clean coordinates with GeoPandas (because some USGS files still use NAD27, bless their hearts)

Model Training

  • Train suitability model with MLlib
  • Cross-validate against historical sighting clusters

Visualization

  • Generate interactive maps showing "Squatch Scores" (0-100)
  • Highlight areas with habitat loss trends

Who Actually Needs This?

  • "As a researcher, I need to identify 10 high-probability survey sites per state before funding runs out."
  • "As a documentary producer, I want to visualize habitat loss over time to make rich people feel guilty."
  • "As a park ranger, I need to know where tourists might get 'surprised' so I can stock up on first-aid kits."

Conservation Tech Spin

This isn't just about finding Bigfoot—it's about protecting what we haven't even discovered yet:

Model Accuracy: 82% vs historical clusters Habitat Loss Alerts: 3-month lead time NGO Adoption Target: 5 pilot programs

By framing it as "preemptive conservation tech", we:

  • Tap into wildlife NGO budgets (bigger than cryptid research grants)
  • Build tools that help known species too
  • Avoid looking like "that Bigfoot person" at conferences (mostly)