Setup & Installation

How to build and run the system

​

Prerequisites

• Docker Desktop (with Docker Compose)
• Git
• Hardware: Minimum 8GB RAM recommended

​

Installation

1. Clone & Navigate

   Clone the repository and move to the ML application directory:

   git clone https://github.com/CollinsNyatundo/demand-forecasting-system.git
   cd demand-forecasting-system/app-ml
   
2. Configure Secrets

   Create your environment file from the template. This is critical for Airflow and MinIO security.

   cp .env.template .env
   

   Edit .env to set your AIRFLOW__SMTP__SMTP_PASSWORD if you want email alerts enabled.

3. Launch Core Services

   Start the MLOps stack (Airflow, Model Server, MinIO):

   docker-compose up --build -d
   

   First launch may take 5-10 minutes to build the optimized Airflow & Model Server images.

4. Launch Dashboard

   The Analytics UI runs as a separate service. Open a new terminal:

   cd ../app-ui
   pip install -r requirements.txt
   python app.py
   
5. Verify Installation

   Check that all containers are healthy:

   docker-compose ps
   

   Then access the services:

   • Airflow Orchestrator: http://localhost:8080 (admin/admin)
   • Analytics Dashboard: http://localhost:8050
   • Prediction API: http://localhost:8000/docs
   • MinIO Storage: http://localhost:9001

​

Troubleshooting

Show Services fail to start?

Check if ports 8080, 8000, or 5432 are occupied by other applications.

docker-compose logs -f

Show Email alerts not working?

Ensure you are using an App Password for Gmail, not your login password. Check .env settings.