Logo Lanfrica

Filimon-H/ethio-med-telegram-analytics

Domain:

natural language processinghealthcare

Record type:

softwareproject
Creator:
Fil
Host:
End-to-end data pipeline analyzing Ethiopian medical Telegram channels. Includes data scraping, warehouse modeling with dbt, image detection (YOLOv8), and FastAPI analytics. Built with Docker, PostgreSQL, and Dagster orchestration." # πŸ₯ ETHIO-MED-TELEGRAM-ANALYTICS **Containerized analytics platform** for Ethiopian medical Telegram data πŸ“¦πŸ‡ͺπŸ‡Ή End-to-end pipeline for NLP/CV tasks: Named Entity Recognition (NER) & YOLOv8 image detection --- ## 🧠 Project Overview Robust **ELT pipeline** ingesting Amharic Telegram messages from Ethiopian medical channels. Processes data for: - 🧠 Amharic NER (Named Entity Recognition) - πŸ‘οΈ YOLOv8 medical object detection - πŸ“Š Business analytics API --- ## πŸ”§ Tech Stack | Layer | Tool | |-------------------|-------------------------------| | 🐘 **Database** | PostgreSQL 14 + pgAdmin | | 🧱 **Transform** | dbt (Data Build Tool) | | πŸ“€ **Ingestion** | Telethon + Custom Python | | πŸ” **API** | FastAPI + Pydantic | | πŸ€– **ML Models** | YOLOv8, HuggingFace Transformers | | βš™οΈ **Orchestration** | Dagster | | 🐳 **Infra** | Docker + Docker Compose | --- ## πŸ“‚ Directory Structure β”œβ”€β”€ data/ # Telegram data β”‚ β”œβ”€β”€ raw/ # Raw JSONs β”‚ └── preprocessed/ # Cleaned data β”œβ”€β”€ ethio_api/ # FastAPI app β”œβ”€β”€ ethio_pipeline/ # Dagster jobs β”œβ”€β”€ telegram_dbt_project/ # dbt models β”œβ”€β”€ scripts/ # Data scripts β”œβ”€β”€ notebooks/ # EDA/ML experiments β”œβ”€β”€ .env # Environment secrets β”œβ”€β”€ docker-compose.yml # Multi-container setup β”œβ”€β”€ requirements.txt # Dependencies └── README.md ## πŸš€ Quickstart Guide ### 1. Clone & Configure ```bash git clone github.com cd ETHIO-MED-TELEGRAM-ANALYTICS cp .env.example .env πŸ“ Edit .env with PostgreSQL and Telegram credentials 2. Launch Containers docker compose up --build 3. Scrape Messages python scripts/scrape_telegram.py --channel @cheMed123 --limit 1000 4. Load Raw Data python scripts/load_raw_json.py --input data/raw/telegram_messages/YYYY-MM-DD/ 5. Run dBT Transformations docker exec -w /app/telegram_dbt_project -it telegram_app dbt build 6. Serve Documentation (Optional) docker exec -w /app/te …