Skip to content

Using a stack of powerful tools to build an End-to-End AutoML pipeline for insurance cross-sell prediction

License

Notifications You must be signed in to change notification settings

aminbenmansour/e2e-insurance-cross-sell-prediction-automl

Repository files navigation

e2e-automl-cross-sell-prediction

⚠️ Disclaimer

⚠️ This is not my idea neither my work, instead of forking this repo, I wanted to follow along step by step with the help of this article.

Special thanks to the original author. mr. Kenneth Leung for his valuable writings and contributions. Please consider supporting him and giving his data science portfolio a look.

introduction

This project aims to make cross-selling more efficient and targeted by building an ML pipeline to identify health insurance customers interested in purchasing additional vehicle insurance.

toolkits overview

  • H2O AutoML

    H2O is an open-source, distributed, and scalable platform that enables users to easily build and productionalize ML models in enterprise environments.

    One of H2O’s key features is H2O AutoML, a service that automates the ML workflow, including the automatic training and tuning of multiple models.

    This automation allows teams to focus on other vital components such as data preprocessing, feature engineering, and model deployment.

  • MLflow

    MLflow is an open-source platform for managing ML lifecycles, including experimentation, deployment, and creation of a central model registry.

    The MLflow Tracking component is an API that logs and loads the parameters, code versions, and artifacts from ML model experiments.

    MLflow also comes with a mlflow.h2o API module that integrates H2O AutoML runs with MLflow tracking.

  • FastAPI

    FastAPI is a fast and highly performant web framework for building APIs in Python. It is designed to be user-friendly, intuitive, and production-ready.

    The purpose of deploying our ML model as a FastAPI endpoint is to readily retrieve prediction results after parsing our test data into the API.

  • Streamlit

    Streamlit is an open-source app framework that turns data scripts into shareable web apps within minutes. It helps create user-friendly front-end interfaces for both technical and non-technical users to interact with.

    The user interface allows us to upload data to the backend pipeline via the FastAPI endpoint and then download predictions with a click of the mouse.

About

Using a stack of powerful tools to build an End-to-End AutoML pipeline for insurance cross-sell prediction

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published