How akchiche-mohamed-aymen/customer-churn-prediction-segmentation-using-machine-learning works
This repository is a Streamlit customer-churn application combining churn-risk prediction, customer segmentation, and AI-generated retention recommendations. The principal workflow is customer input in the dashboard, API prediction, and recommendation display. A separate preparation and model-training workflow creates the artifacts used at runtime. Source excerpts support the model and recommendation internals; the README documents the intended setup sequence, while some UI call details are unsampled.
Main entry points: api.py Prediction API · unsupervised.py Segmentation Training · prepped_data.csv Prepared Data · get_recom.py Recommendation Generator
Components 16
Data and Training
- Churn Dataset
Bank Customer Churn Prediction.csv - Data Preparation
prep_data.py - Prepared Data
prepped_data.csv - Cluster Selection
choosing_best_k.py - Selected K
best_k.json - Segmentation Training
unsupervised.py - Classification Data
data_for classification.csv - Churn Model Training
supervised.py - K-Means Model
kmeans_model.pkl - Random Forest Model
random_forest_clf.pkl
Prediction Runtime
- Prediction API
api.py
Recommendations
- Recommendation Generator
get_recom.py - Response Schemas
schemas.py
Dashboard
- Streamlit Dashboard
UI.py
Other components
- Business User
- Google Gemini
Connections 18
- Business User → Streamlit Dashboard: enters customer data
README.md - Streamlit Dashboard → Prediction API: requests prediction
UI.py - Streamlit Dashboard → Recommendation Generator: imports generator
UI.py - Prediction API → K-Means Model: loads and predicts
api.py - Prediction API → Random Forest Model: loads and predicts
api.py - Prediction API → Classification Data: reads feature columns
api.py - Recommendation Generator → Response Schemas: formats response
get_recom.py - Recommendation Generator → Google Gemini: requests recommendations
get_recom.py - Churn Dataset → Data Preparation: provides source data
prep_data.py - Data Preparation → Prepared Data: writes prepared data
prep_data.py - Prepared Data → Cluster Selection: supplies clustering data
choosing_best_k.py - Cluster Selection → Selected K: writes selected K
choosing_best_k.py - Prepared Data → Segmentation Training: supplies training data
unsupervised.py - Selected K → Segmentation Training: configures cluster count
unsupervised.py - Segmentation Training → Classification Data: writes segmented data
unsupervised.py - Segmentation Training → K-Means Model: saves pipeline
unsupervised.py - Classification Data → Churn Model Training: supplies training data
supervised.py - Churn Model Training → Random Forest Model: saves classifier
supervised.py