An open statistical archive Tbilisi, Georgia · Updated August 19, 2026

EconGE.

The Georgian economic record, collected nightly.

An automated pipeline that gathers, normalizes, and publishes economic statistics from Georgia's official sources — GeoStat, the National Bank, the Ministry of Finance, the World Bank, and the IMF — into one browsable archive with a structured public API.

Browse the data → Public API Pipeline UI
3.4M
Observations
127,547
Datasets
5
Sources
Aug 19
Last updated

The Archive

Five official sources, one record

Each source is scraped on its own schedule, parsed into structured datasets, and loaded into a normalized PostgreSQL schema.

IMF — Georgia https://www.imf.org/en/Countries/GEO
9 datasets 328 obs 1994–2031 api
Ministry of Finance of Georgia https://www.mof.ge
1,060 datasets 21.2K obs 2002–2026 html scrape
National Bank of Georgia https://nbg.gov.ge
42 datasets 20.5K obs 2016–2026 api+html scrape
National Statistics Office of Georgia https://www.geostat.ge
11,683 datasets 3.3M obs 1903–2100 html scrape
World Bank — Georgia https://data.worldbank.org/country/georgia
8 datasets 366 obs 1960–2025 api

Why this exists

The record was scattered.
Now it isn't.

Georgian economic data is published across multiple government websites in inconsistent formats — Excel spreadsheets, PDFs, and fragmented APIs. Finding a historical time series, or comparing indicators across sources, is painful.

EconGE automates the collection, normalization, and storage of this data into a single PostgreSQL database with a clean, browsable interface on top. It runs on a schedule, so the record stays fresh without manual intervention.

It is both a practical tool and a working demonstration of end-to-end data engineering: ingestion, transformation, storage, orchestration, and delivery.

In practice

  1. 01 Automated daily scraping of official sources
  2. 02 Normalized schema across all data providers
  3. 03 Browsable portal with search and charts
  4. 04 Time-series visualization on every dataset
  5. 05 Dagster-managed pipeline with monitoring
  6. 06 Dockerized deployment on a single VPS
  7. 07 JSON API for programmatic access
  8. 08 Incremental loading — only new data is fetched

Colophon

How it's built

A modern data stack running on a single VPS — no cloud vendor lock-in.

Orchestration Dagster · Dagster Daemon · Cron Schedules
Collection Python · httpx · openpyxl · REST APIs
Storage PostgreSQL 16 · Normalized Schema · Docker Volumes
Delivery FastAPI · Nginx · Docker Compose · Let's Encrypt
Fig. 1 — Pipeline architecture 
  ┌──────────────┐     ┌──────────────┐     ┌──────────────┐
  │  GeoStat API │     │   NBG API    │     │  Excel Files │
  └──────┬───────┘     └──────┬───────┘     └──────┬───────┘
         │                    │                    │
         └────────────┬───────┴────────────────────┘
                      │
              ┌───────▼────────┐
              │    Dagster      │
              │  (Orchestrator) │
              └───────┬────────┘
                      │
              ┌───────▼────────┐
              │  Python Assets  │
              │  (Extract +     │
              │   Transform)    │
              └───────┬────────┘
                      │
              ┌───────▼────────┐
              │  PostgreSQL 16  │
              │  (econge schema)│
              └───────┬────────┘
                      │
         ┌────────────┴────────────┐
         │                         │
  ┌──────▼───────┐      ┌─────────▼────────┐
  │  Data Portal │      │  Dagster Web UI  │
  │  (FastAPI)   │      │  (Pipeline Ops)  │
  └──────────────┘      └──────────────────┘

Read the record.

Browse two thousand series, chart any of them, pull them through the API — or open the pipeline and watch the machinery run.

Data portal → Pipeline UI Back to main site