POI Activity Index — Geospatial Data Pipeline & DeFi on Flare

A data pipeline that ingests open-source geospatial and social data, computes a deterministic commercial activity index across 50+ London locations, and feeds it on-chain to power parametric insurance payouts and yield adjustments on the Flare blockchain.

Live Demo →

---

What It Does

Most commercial risk assessment relies on lagging indicators — quarterly revenue reports, annual foot traffic surveys. This system computes a real-time activity index from publicly available data, creating a forward-looking revenue proxy that can trigger automated financial logic.

The pipeline has two layers:

Off-chain: Geospatial data pipeline Ingests, cleans, and normalises multi-source data to produce a standardised activity score per location.

Source	Signal	Why it matters
OpenStreetMap	POI density, commercial land use	Structural capacity of an area
Floor area data	Physical commercial footprint	Scale of commercial activity
Ratings data	Consumer activity and sentiment	Demand-side signal
Event data	Temporal activity spikes	Captures short-term surges

On-chain: DeFi smart contracts (Flare Coston2 testnet) Activity index outputs are attested via the Flare Data Connector (FDC) and consumed by Solidity contracts that execute parametric insurance payouts and yield rate adjustments — no manual claims process, no trusted intermediary.

---

Architecture

Data Sources (OSM, ratings, events, floor area)
        |
        v
  Data Pipeline         <- Ingest, clean, normalise
  (Python)
        |
        v
  Activity Index        <- Deterministic, explainable score
  Computation
        |
        v
  Flare Data            <- Attest off-chain data on-chain
  Connector
        |
        v
  Smart Contracts       <- Parametric insurance + yield logic
  (Solidity)
        |
        v
  Frontend              <- Map UI, wallet flows, settlement
  (Next.js)

---

Key Design Decisions

• No black-box ML — the activity index is deterministic and explainable, making it auditable for financial applications
• No Google APIs — built entirely on open-source data (OpenStreetMap, Mapbox, Reddit, event APIs)
• Flare-native — uses FDC for attested Web2 data rather than a centralised oracle, with FTSO price feeds for on-chain settlement

---

Tech Stack

• Data pipeline: Python, Pandas, OpenStreetMap API, Mapbox
• Smart contracts: Solidity, Foundry, Flare Coston2 testnet
• Frontend: Next.js, wagmi, viem
• On-chain infrastructure: Flare Data Connector (FDC), FTSO oracle

---

My Contribution

• Engineered the data ingestion and cleaning pipeline across all four sources
• Designed the normalisation and weighting logic for the composite activity index
• Built data visualisations to present index scores for downstream analysis

---

Repo Structure

├── data/          # Data pipeline & geospatial processing
├── contracts/     # Flare smart contracts (Foundry)
└── frontend/      # Next.js + wagmi (Flare Coston2)

See each folder's README for setup details.

---

Local Setup

• data/ — Copy data/.env.example to data/.env if using APIs that need keys
• contracts/ — Copy contracts/.env.example to contracts/.env, set PRIVATE_KEY. Use Flare Coston2 RPC
• frontend/ — Copy frontend/.env.local.example to frontend/.env.local, paste contract addresses after deploying. Get C2FLR from the Flare Coston2 Faucet

License

MIT