GeoInvest

Geospatial real estate analytics for Baku, Azerbaijan. Aggregates property listings onto Uber's H3 hexagonal grid to surface price trends, demand density, and market shifts across the city — by district, resolution, and time period.

What it does

• Maps median price per m² and listing density across H3 hexagons at resolutions H6–H8
• Dual analysis modes: administrative boundary (geom) and pure H3 grid
• Period-over-period trend tracking with automatic previous-period comparison
• Adjustable outlier filter to remove low-sample hexagons from analysis
• Color-by control: switch between price heatmap and density heatmap instantly

Stack

Layer	Tech
Frontend	Next.js 14, deck.gl H3HexagonLayer, MapLibre GL, React Query
Backend	FastAPI, SQLAlchemy async, PostgreSQL 16 + PostGIS
Cache	Redis (120s TTL)
Data pipeline	Python + psycopg2, H3 spatial aggregation via PostgreSQL
Infrastructure	Docker Compose

Project structure

GeoInvest/
├── app/                    # Main application
│   ├── backend/            # FastAPI + SQLAlchemy
│   ├── frontend/           # Next.js + deck.gl
│   └── docker-compose.yml
├── pipeline/               # Data pipeline
│   ├── generate_analysis.py   # Pulls from source DB, writes data/h3_analysis.csv
│   └── export_raw_data.py     # Raw export utility
├── data/                   # Data files
│   ├── h3_analysis.csv        # Pre-aggregated H3 analytics (committed)
│   ├── baku_districts.json    # District boundary GeoJSON
│   └── baku_rayons.geojson
└── scripts/                # Exploration & utility scripts

Quick start

# 1. Copy env files
cp .env.example .env
cp app/backend/.env.example app/backend/.env
cp app/frontend/.env.example app/frontend/.env.local

# 2. Start services
cd app
docker compose up -d db redis
docker compose run --rm backend python scripts/ingest_csv.py
docker compose up --build backend frontend

# 3. Open http://localhost:3000

Regenerating data from source

Set environment variables and run the pipeline:

export SOURCE_DB_HOST=your_host
export SOURCE_DB_USER=your_user
export SOURCE_DB_PASSWORD=your_password

python pipeline/generate_analysis.py
cd app && docker compose run --rm backend python scripts/ingest_csv.py

Environment variables

Backend (app/backend/.env):

Variable	Default	Description
DATABASE_URL	postgresql+asyncpg://h3:h3@localhost:5433/h3_analytics	Analytics DB
REDIS_URL	redis://localhost:6379/0	Cache
RAW_DATA_CSV_PATH	data/h3_analysis.csv	Input data
BOUNDARIES_JSON_PATH	data/baku_districts.json	District boundaries
CACHE_TTL_SECONDS	120	API cache TTL

Pipeline (environment variables):

Variable	Description
SOURCE_DB_HOST	Source database host
SOURCE_DB_USER	Source database user
SOURCE_DB_PASSWORD	Source database password
SOURCE_DB_NAME	Source database name (default: source_dev_db)