How morshadul/customer-churn-prediction works
This repository is a Streamlit customer-churn prediction app. A user enters customer attributes; the app encodes categorical values, scales the assembled inputs, invokes a saved neural-network model, and displays a churn probability and classification. The graph focuses on this inference workflow. The README gives no further capability details, and the notebooks and training workflow were not sampled, so they are not mapped.
Main entry points: app.py Input preparation · model.h5 Churn model · label_encoder_gender.pkl Gender encoder · one_hot_encoder_geo.pkl Geography encoder
Components 8
User interface
- Customer inputs
app.py - Prediction display
app.py
Prediction
- Gender encoder
label_encoder_gender.pkl - Geography encoder
one_hot_encoder_geo.pkl - Input preparation
app.py - Feature scaler
scaler.pkl - Churn model
model.h5
Other components
- User
Connections 8
- User → Customer inputs: enters attributes
- Customer inputs → Input preparation: passes inputs
- Input preparation → Gender encoder: encodes gender
- Input preparation → Geography encoder: encodes geography
- Input preparation → Feature scaler: scales features
- Input preparation → Churn model: predicts churn
- Churn model → Prediction display: returns probability
- Prediction display → User: shows prediction