ML-Ops Playground

A machine learning operations (MLOps) project demonstrating MLflow integration for experiment tracking and model management using the Iris dataset.

Project Structure

ml-ops-playground/
├── .dockerignore
├── .gitignore
├── Dockerfile.mlflow
├── Dockerfile.training
├── README.md
├── compose.yml
├── requirements.txt
└── training.py

Prerequisites

• Docker and Docker Compose
• Python 3.9+
• MLflow
• scikit-learn

Quick Start

1. Clone the repository:

git clone https://github.com/yourusername/ml-ops-playground.git
cd ml-ops-playground

2. Start the MLflow server and training service:

docker compose up --build

The services will:

• Start MLflow server on http://localhost:8080
• Execute the training script automatically

Project Components

MLflow Server

• Tracks experiments and metrics
• Stores model artifacts
• Accessible via web UI at http://localhost:8080

Training Pipeline

• Uses scikit-learn's Iris dataset
• Implements Logistic Regression model
• Logs metrics, parameters, and model artifacts to MLflow

Model Details

• Algorithm: Logistic Regression
• Dataset: Iris Classification
• Metrics: Accuracy
• Parameters:
  • solver: lbfgs
  • max_iter: 1000
  • multi_class: auto
  • random_state: 8888

Environment Variables

• MLFLOW_TRACKING_URI: URI for MLflow tracking server (default: http://mlflow:8080)

Development

To modify the training pipeline:

1. Update training.py
2. Rebuild and run containers:

docker compose up -d --build

License

[Your License Here]

Contributing

1. Fork the repository
2. Create your feature branch
3. Commit your changes
4. Push to the branch
5. Create a new Pull Request