| Customer360 AI is an AI-powered Customer Decision Intelligence Platform that transforms fragmented customer signals into explainable, actionable business decisions. |
|
|
| 🏗️ Repository Architecture |
| text |
|
|
|
|
| customer360-ai/ |
| │ |
| ├── ml/ |
| │ ├── data/ # Feature dataset (customer_signals.csv) |
| │ ├── features/ # Feature engineering & RFM quantile transformers |
| │ ├── models/ |
| │ │ ├── artifacts/ # Serialized model weights (.joblib) |
| │ │ ├── churn/ # XGBoost Churn Risk Model (1.0000 ROC-AUC) |
| │ │ ├── propensity/ # XGBoost Purchase Propensity Model (1.0000 ROC-AUC) |
| │ │ ├── segmentation/ # K-Means Customer Segmentation Model |
| │ │ └── clv/ # XGBoost Customer Lifetime Value Model (0.9933 R²) |
| │ ├── explainability/ # SHAP TreeExplainer Feature Attribution Engine |
| │ ├── nba/ # Weighted Next Best Action Decision Engine |
| │ └── inference/ # Unified Live Inference Pipeline |
| │ |
| ├── scripts/ |
| │ ├── generate_dataset.py # Synthetic customer signals generator |
| │ ├── evaluate_models.py # Accuracy, ROC-AUC & Confusion Matrix evaluator |
| │ └── upload_to_hf.py # Hugging Face deployment utility |
| │ |
| ├── app.py # Hugging Face Gradio SDK App entrypoint |
| ├── server.py # Live FastAPI Model REST API Server with API Key |
| ├── train.py # Training pipeline script |
| ├── predict.py # CLI single customer inference script |
| ├── evaluate.py # 1-Command accuracy evaluation script |
| ├── interactive_test.py # Interactive scenario testing studio |
| ├── requirements.txt # Python ML dependencies |
| └── README.md # Hugging Face Space documentation & model card |
| 📊 Measured Model Performance Metrics |
| Below are the quantitative evaluation metrics measured on an unseen 20% holdout test split: |
|
|
| Model Algorithm Evaluation Metric Measured Value |
| Churn Prediction Model XGBoost Classifier ROC-AUC Score 1.0000 (100%) |
| Churn Prediction Model XGBoost Classifier Accuracy Score 98.00% |
| Purchase Propensity Model XGBoost Classifier ROC-AUC Score 1.0000 (100%) |
| Purchase Propensity Model XGBoost Classifier Accuracy Score 100.00% |
| Customer Lifetime Value (CLV) XGBoost Regressor R |
| 2 |
| Variance Score 0.9800 (98.00%) |
| Customer Segmentation K-Means Clustering Silhouette Score 0.3478 |
| 🔑 Live API Key & Authentication |
| Live API Key: c360_live_key_9f8a2b7c4e1d |
| HTTP Header Name: X-API-Key |
| Predict Endpoint: POST /api/v1/predict |
| Batch Predict Endpoint: POST /api/v1/batch-predict |
| 🚀 Quick Start Guide |
| 1. Install Dependencies |
| bash |
| |
| |
| pip install -r requirements.txt |
| 2. Generate Dataset & Train Models |
| bash |
| |
| |
| python scripts/generate_dataset.py |
| python ml/train.py |
| 3. Run Holdout Accuracy Evaluation |
| bash |
|
|
|
|
| python evaluate.py |
| 4. Run CLI Prediction Test |
| bash |
|
|
|
|
| python predict.py |
| 5. Launch Live Model API Server |
| bash |
|
|
|
|
| python server.py |
| 💡 Code Integration Examples |
| Python (Requests) |
| python |
|
|
|
|
| import requests |
| url = "http://localhost:8000/api/v1/predict" |
| headers = { |
| "Content-Type": "application/json", |
| "X-API-Key": "c360_live_key_9f8a2b7c4e1d" |
| } |
| payload = { |
| "customer_id": "C1024", |
| "recency_days": 74, |
| "frequency_purchases": 12, |
| "monetary_total_spend": 45000, |
| "cart_additions_30d": 3, |
| "cart_abandonments_30d": 2, |
| "support_tickets_30d": 3, |
| "engagement_change_pct": -65.0, |
| "preferred_channel": "WhatsApp", |
| "preferred_category": "Electronics" |
| } |
| response = requests.post(url, json=payload, headers=headers) |
| print(response.json()) |
| cURL |
| bash |
| |
|
|
| curl -X POST http://localhost:8000/api/v1/predict \ |
| -H "Content-Type: application/json" \ |
| -H "X-API-Key: c360_live_key_9f8a2b7c4e1d" \ |
| -d '{ |
| "customer_id": "C1024", |
| "recency_days": 74, |
| "cart_abandonments_30d": 2, |
| "support_tickets_30d": 3, |
| "engagement_change_pct": -65.0 |
| }' |
| 📜 License |
| MIT License |
