Customer Churn Prediction System

Project Overview

This project was developed for Task 2 of the Machine Learning Internship at Future Interns. The system identifies customers likely to stop using a service, allowing businesses to take proactive retention measures.

Tech Stack

The versions below are optimized for 2026 deployment environments, specifically addressing library deprecations and Streamlit Cloud compatibility.

1. UI: Streamlit (v1.52.2)
2. Model: FastAI Tabular Learner (v2.8.6)
3. Interpretation: SHAP (v0.49.1)
4. Data Handling: Pandas (v2.3.3) & NumPy (v2.0.2)
5. Environment Fixes: Added ipython for fastprogress support on Linux servers.
6. Training Environment: Kaggle

Repository Structure

1. streamlit/app.py: The main Streamlit web application.
2. streamlit/churn_model.pkl: The exported FastAI model (uses relative paths for cloud hosting).
3. requirements.txt: List of dependencies for cloud deployment.
4. notebooks/: Contains the original Kaggle training notebook (.ipynb).
5. data/: Sample dataset used for training and testing.

Model & Performance

The model was trained on the Telco Customer Churn dataset.

1. Type: Deep Learning (Tabular Learner).
2. Interpretability: Uses SHAP to explain feature importance and prediction drivers.
3. Key Features: Contract type, Tenure, Monthly Charges, and Internet Service.

How to Run Locally

1. Clone the repository:

   git clone https://github.com/AnoMi-1/FUTURE_ML_02.git
   cd FUTURE_ML_02
   

2. Install dependencies: pip install -r requirements.txt

3. Run the application: streamlit run streamlit/app.py