README.md

metadata

title: ECL Risk Analyzer
emoji: 💹
colorFrom: red
colorTo: red
sdk: docker
app_port: 8501
tags:
  - streamlit
  - credit-risk
  - gemini
  - finance
pinned: false
short_description: AI ECL analyzer using Streamlit and Gemini.

🧮 ECL Risk Analyzer

ECL Risk Analyzer is a Streamlit-based web app that automates Expected Credit Loss (ECL) analysis and provides AI-driven recommendations on credit decisions.

🔍 Features

• Upload loan datasets (CSV format).
• Compute key credit metrics:
  • PD – Probability of Default
  • LGD – Loss Given Default
  • EAD – Exposure at Default
  • ECL – Expected Credit Loss
• Visualize segment-level performance with interactive charts.
• Get AI recommendations (via Gemini 2.5 Flash Lite) to:
  • Increase interest
  • Reduce disbursement
  • Maintain current policy
• Role-based access:
  • Analyst: Run and review segment analyses.
  • CRO: Audit all analyst reports and view trends.

🧠 Tech Stack

• Frontend: Streamlit
• AI Engine: Gemini API
• Database: SQLite (via SQLAlchemy)
• Hosting: Hugging Face Spaces (Docker SDK)

🚀 Quick Start

1. Upload your loan dataset (CSV).
2. Select a segmentation column (e.g., loan intent, area, or occupation).
3. View computed ECL metrics and charts.
4. Choose a segment to get AI-based recommendations.

📊 Example Output

Segment	PD	LGD	EAD	ECL	Action
Medical	0.278	0.255	78M	5.5M	Increase Interest

🛠️ Customization

Modify /src/streamlit_app.py to:

• Adjust prompts or logic.
• Change chart styles.
• Extend user roles or security.