Financial Twin

What I Learned

Transaction Data Extraction: Using Python scripts and AI models to extract structured transaction data from PDFs. Classification & Aggregation: Implementing machine learning (Gemini) for transaction categorization and grouping expenses into meaningful subcategories. LSTM for Forecasting: Training a deep learning model to predict future spending trends. UI Implementation: Developing an interactive UI with Streamlit/Gradio for dynamic budget simulation. How I Built It

Data Extraction: Extracted transaction details from PDF bank statements using Extraction.py and Gemini AI. Classification & Aggregation: Classified transactions using AI and grouped them by category for monthly analysis. Forecasting with LSTM: Trained an LSTM model to predict spending patterns for the next 6-12 months. Expense Analysis & Savings Suggestions: Provided insights based on spending habits and AI-driven recommendations. Interactive UI: Developed a user-friendly interface allowing users to simulate financial scenarios dynamically. Challenges Faced

Data Quality Issues: Extracting structured data from PDFs required handling formatting inconsistencies.

Built With

• css
• flask
• html
• lstm
• python
• pytorch
• react
• scikit-learn
• tensor