/churn-prediction

This application uses Machine Learning to predict churning customers of a bank while providing useful evaluation metrics and KPIs.

Primary LanguagePythonMIT LicenseMIT

Explainable AI for Churn Prediction

This project was part of Becode AI Bootcamp

App deployed on Streamlit and Heroku

Table of contents

Description
Installation
Usage

Description

This application aims at predicting churning customers from a bank. The original dataset as well as a description of the project can be found on Kaggle

A particular feature of this otherwise extremely clean dataset is that it is highly unbalanced; only 16.07% of customers are flagged as "attrited".

The streamlit application is divided into 2 sections:

  1. Exploratory Data Visualization. The goal of this section is to display important features of the dataset, and to be able to visualize data points in a 3D scatter plot with the binary label (attrited / existing).

  2. Prediction. At the top of this section one can find KPIs regarding the chosen model from the sidebar, including the parameters of the trained model. Users can also display a more advanced dashboard with single predictions from the testing set, and the SHapley Additive exPlanation (SHAP) relative to this particular model's decision.

Installation

  1. Clone the repository:
git clone https://github.com/CorentinChanet/churn-prediction
  1. Install the required libraries:
pip install -r requirements.txt

Usage

To start the program on your local machine:

streamlit run streamlit_app.py