to-olx/fintech-ml-app

🎯 Personal Finance ML Application for South African financial data. Features advanced transaction categorization (93.3% accuracy), SARIMAX forecasting, multi-debt optimization, and interactive Streamlit dashboards. Built with domain-driven architecture. # 💰 Personal Finance ML Application A sophisticated Personal Finance Machine Learning Application designed specifically for South African financial data. Process multiple bank CSV files, perform advanced transaction categorization, implement ML forecasting, and optimize debt strategies. ## 🎯 Key Features - **📊 Multi-Bank Data Integration**: Processes 4+ different South African bank CSV formats - **🏷️ Smart Categorization**: 93.3% accuracy with SA-specific transaction patterns - **🤖 ML Forecasting**: SARIMAX time series models for cashflow prediction - **💳 Debt Optimization**: Multi-debt portfolio optimization with avalanche/snowball strategies - **📈 Interactive Dashboards**: 3 different Streamlit interfaces + CLI - **📋 Budget Analysis**: 3-year historical budget vs actual spending analysis - **📄 Export & Reporting**: PDF reports, CSV exports, implementation schedules ## 🚀 Quick Start ### Prerequisites - Python 3.11+ - uv package manager ### Installation ```bash # Clone the repository git clone github.com cd fintech-ml-app # Install dependencies uv sync ``` ### Usage #### 1. CLI Interface ```bash # Run ML pipeline with default settings uv run fintech # Custom forecast horizon uv run fintech --horizon-days 60 --data-dir ./data ``` #### 2. Web Dashboards ```bash # Main comprehensive dashboard uv run streamlit run fintech/dashboard.py --server.port 8501 # Debt optimization dashboard uv run streamlit run fintech/debt_dashboard.py --server.port 8502 # Simple clean dashboard uv run streamlit run fintech/simple_dashboard.py --server.port 8503 ``` ## 📊 Current Performance - **5,657 transactions** processed from multiple bank accounts - **93.3% categorization accuracy** with SA market-specific rules - **341 inter-account transfers** automatically detected and excluded - **3+ years** of financial data (2022-2025) - **SARIMAX ML model** with validation RMSE: 8,962.54 ## 🏗️ Architecture ### Domain-Driven Design - **Domain Layer …