MLOPS with MLflow

This is a sample project that demonstrates how to use MLflow to create an MLOps pipeline for machine learning models. The project includes two options for installation: local and Docker.

Requirements

• Python 3.x
• Jupyter Notebook
• MLflow
• Docker (optional)
• Nginx (optional)

Installation

Local Installation

Clone the repository and navigate to the directory:

git clone https://github.com/Adeyeha/MLOps-with-MLflow.git
cd <repo-directory>

Create a virtual environment and activate it:

python3 -m venv env
source env/bin/activate

Follow Instructions on Notebook Getting-Started-with-Mlflow.ipynb in experiments directory

Docker Installation

Clone the repository and navigate to the directory:

git clone https://github.com/Adeyeha/MLOps-with-MLflow.git
cd <repo-directory>

Run Docker Compose:

docker-compose build
docker-compose up

How to use

Start the Jupyter Notebook server:

jupyter notebook

Navigate to the notebook Getting-Started-with-Mlflow.ipynb and run the cells to train and deploy the machine learning model.

To track the model using MLflow, open a separate terminal window and navigate to the project directory. Then, start the MLflow tracking server:

mlflow server --host 0.0.0.0

To view the MLflow tracking UI, open a web browser and navigate to http://localhost:5000.

Contents

The Getting-Started-with-Mlflow.ipynb notebook contains the following sections:

• Data preparation: Preprocessing the data and splitting it into training and testing datasets.
• Training the model: Training a machine learning model using scikit-learn and saving the model in MLflow.
• Tracking the model: Starting an MLflow experiment and logging the model parameters, metrics, and artifacts.
• Deploying the model: Deploying the model as a REST API using MLflow.
• Serving the model: Testing the deployed model by sending requests to the API.

Contributing

If you want to contribute to this project, please create a pull request with a detailed description of your changes.

Author

Temitope Adeyeha