Update README.md
Browse files
README.md
CHANGED
|
@@ -1,20 +1,53 @@
|
|
| 1 |
---
|
| 2 |
title: Fraud Guard Intelligence
|
| 3 |
-
emoji:
|
| 4 |
-
colorFrom:
|
| 5 |
-
colorTo:
|
| 6 |
-
sdk:
|
| 7 |
-
|
| 8 |
-
|
| 9 |
-
|
| 10 |
-
pinned: false
|
| 11 |
-
short_description: Streamlit template space
|
| 12 |
license: mit
|
| 13 |
---
|
| 14 |
|
| 15 |
-
#
|
| 16 |
|
| 17 |
-
|
| 18 |
|
| 19 |
-
|
| 20 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
| 2 |
title: Fraud Guard Intelligence
|
| 3 |
+
emoji: 🛡️
|
| 4 |
+
colorFrom: blue
|
| 5 |
+
colorTo: indigo
|
| 6 |
+
sdk: streamlit
|
| 7 |
+
sdk_version: 1.32.0
|
| 8 |
+
app_file: app.py
|
| 9 |
+
pinned: true
|
|
|
|
|
|
|
| 10 |
license: mit
|
| 11 |
---
|
| 12 |
|
| 13 |
+
# 💳 Fraud Guard Intelligence: Adaptive MLOps System
|
| 14 |
|
| 15 |
+
This repository hosts the interactive frontend for a production-ready Fraud Detection system. The architecture is designed to handle non-static fraud patterns through real-time inference, explainable AI, and a human-in-the-loop retraining pipeline.
|
| 16 |
|
| 17 |
+
## 🚀 System Architecture
|
| 18 |
+
This is a **Microservices-based Deployment**:
|
| 19 |
+
1. **Frontend (This Space):** Streamlit UI for transaction input and SHAP-based risk visualization.
|
| 20 |
+
2. **Backend (Render/Railway):** FastAPI server handling high-concurrency inference requests.
|
| 21 |
+
3. **Data Layer (Neon DB):** PostgreSQL cloud database logging real-time telemetry for drift analysis.
|
| 22 |
+
4. **Experiment Tracking (DagsHub/MLflow):** Versioned model registry and performance tracking.
|
| 23 |
+
|
| 24 |
+
## 🧠 Key Features
|
| 25 |
+
- **Explainable AI:** Uses SHAP Waterfall plots to visualize why a specific transaction was flagged.
|
| 26 |
+
- **Dynamic Sensitivity:** Stakeholders can adjust the "AI Sensitivity" slider to balance False Positives vs. False Negatives.
|
| 27 |
+
- **Drift Monitoring:** Integrated with Evidently AI to detect statistical shifts in incoming data distribution.
|
| 28 |
+
- **Auto-Retraining:** Trigger a model refresh directly from the UI when enough human-verified data is collected.
|
| 29 |
+
|
| 30 |
+
## 🛠️ Tech Stack
|
| 31 |
+
- **Languages:** Python
|
| 32 |
+
- **ML Frameworks:** Scikit-Learn, XGBoost (v2.0.3)
|
| 33 |
+
- **APIs:** FastAPI, Uvicorn
|
| 34 |
+
- **Tracking:** MLflow, DagsHub
|
| 35 |
+
- **Database:** SQLAlchemy, PostgreSQL (Neon)
|
| 36 |
+
- **Monitoring:** Evidently AI, SHAP
|
| 37 |
+
|
| 38 |
+
## 🔐 Environment Setup
|
| 39 |
+
To run this project locally or on your own Space, ensure the following **Secrets** are configured:
|
| 40 |
+
- `DATABASE_URL`: Cloud PostgreSQL connection string.
|
| 41 |
+
- `API_URL`: The URL of your deployed FastAPI backend.
|
| 42 |
+
- `DAGSHUB_USER_TOKEN`: For MLflow logging.
|
| 43 |
+
- `MLFLOW_TRACKING_URI`: DagsHub MLflow remote URI.
|
| 44 |
+
|
| 45 |
+
## 📂 Project Structure
|
| 46 |
+
- `app/`: Contains the Streamlit dashboard logic.
|
| 47 |
+
- `src/`: Modular code for pipelines (Predict, Train, Drift).
|
| 48 |
+
- `artifacts/`: Serialized model files (`.pkl`) and scalers.
|
| 49 |
+
- `app.py`: The entry point for Hugging Face deployment.
|
| 50 |
+
|
| 51 |
+
---
|
| 52 |
+
**Author:** Mohit Parmar
|
| 53 |
+
**GitHub:** [MohitParmar78](https://github.com/MohitParmar78)
|