Loan Collection Chatbot

The Loan Collection Chatbot is an intelligent, empathetic, and compliant solution designed to streamline loan recovery. It combines predictive modeling, persona-based strategies, and advanced conversational AI to deliver personalized, effective interactions. By leveraging a customer’s past financial history, behavioral data, and real-time intent signals, it adapts to each individual’s needs.

Built with Streamlit and Python, it is designed to be run easily and reliably.

• Predictive Default Model: An XGBoost model that accurately predicts loan default probabilities.
• Persona Identification: Uses KMeans clustering to segment customers into actionable personas (e.g., "Willing but Struggling," "High-Risk Avoider").
• Dynamic Conversational AI: Powered by the Groq LLM to generate adaptive, human-like, and context-aware responses.
• RAG-Powered Compliance: A Retrieval-Augmented Generation (RAG) system ensures all conversations adhere to internal policies and regulatory guidelines from a knowledge base.
• Real-time Analytics: The UI displays live metrics on the conversation's health, including:
  • Customer Sentiment & Intent
  • Negative Interaction Strike Count
  • Dynamic LLM-generated Strategy for each turn
  • A Cumulative Success Score to grade the interaction.
• Containerized & Portable: Packaged with a Dockerfile and docker-compose.yml for simple, one-command deployment.

Running this project with Docker is the simplest way to get started. It handles all dependencies and configurations for you.

• Git: To clone the repository.
• Docker Desktop: Make sure it is installed and running on your system.

Open your terminal and clone the project to your local machine.

git clone https://github.com/your-username/loan-collection-chatbot.git
cd loan-collection-chatbot

The chatbot requires an API key from Groq to function.

• In the project's root directory, find the config.py file.
• Open it and replace the placeholder with your actual Groq API key:

# config.py

# Replace "gsk_..." with your key
GROQ_API_KEY = "gsk_YourActualGroqApiKeyHere"

# ... other settings

With Docker Desktop running, use Docker Compose to build the image and start the application with a single command.

docker-compose up --build

• This command reads the docker-compose.yml file.
• The --build flag tells Docker to build the image from the Dockerfile if it doesn't exist or if the files have changed.
• The process may take a few minutes the first time as it downloads the base image and installs all Python packages.

Once the container is running, open your web browser and navigate to:

http://localhost:8501

To stop the container, return to your terminal and press CTRL + C. To clean up and remove the container completely, run:

docker-compose down

loan-collection-chatbot/
│
├── app.py                   # Main Streamlit application file
├── config.py                # Configuration for API keys and paths
│
├── src/                     # Source code for the chatbot logic
│   ├── chatbot_engine.py    # The core IntegratedChatbot class
│   ├── rag_handler.py       # Functions for RAG setup and vectorstores
│   └── strategy_playbook.py # Persona-based conversation strategies
│
├── scripts/                 # Standalone scripts for model training
│   ├── train_persona_model.py
│   └── train_predictive_model.py
│
├── data/                    # Dataset file
│   └── Analytics_loan_collection_dataset.csv
│
├── assets/                  # Knowledge base files for RAG
│   ├── rbi_guidelines.pdf
│   └── ...
│
├── models/                  # Saved model artifacts (.pkl files)
│
├── Dockerfile               # Instructions to build the Docker image
├── docker-compose.yml       # Orchestration for running the container
├── requirements.txt         # Python package dependencies
└── README.md                # This file

# config.py

# --- API Keys ---
# IMPORTANT: It is recommended to use Streamlit secrets for deployment.
# For local development, you can set your API key here.
GROQ_API_KEY = "YOUR_API_KEY"

# --- Model Configuration ---
GROQ_MODEL_NAME = "openai/gpt-oss-120b"

# --- File Paths ---
DATA_FILE_PATH = 'data/Analytics_loan_collection_dataset.csv'
RAG_PDF_PATHS = [
    'assets/rbi_guidelines.pdf',
    'assets/best-practices-in-collections-strategies.pdf'
]

# --- Asset Paths (for saving/loading trained models) ---
PREDICTIVE_MODEL_ASSETS_PATH = 'models/predictive_assets.pkl'
PERSONA_MODEL_ASSETS_PATH = 'models/persona_assets.pkl'

# --- Chatbot Settings ---
MAX_NEGATIVE_STRIKES = 3

Pull requests are welcome. For major changes, please open an issue first to discuss what you would like to change. For more detailed insights and code, visit the Google Collab - Link