Fund Life

Homepage
Tailored financial plan
Financial analysis

Inspiration

Managing personal finances and investments can be challenging due to the variety of account types (TFSA, RRSP, FHSA, unregistered) and lack of personalized advice. Fund Life aims to analyze personal financial data and provide tailored investment recommendations to maximize returns, reduce debt, and optimize overall financial health.

What It Does

Fund Life helps users:

Receive personalized investment recommendations across multiple account types.
Input their financial data (income, savings, debt, goals) and get real-time guidance.
Understand suggested investment allocations through a user-friendly interface.

How It Was Built

Languages & Frameworks:

Python, Node.js, .NET (C#), MySQL

Machine Learning Model:

Random Forest Regression trained on 10K+ synthetic records
Libraries: scikit-learn, TensorFlow, PyTorch, Joblib
Preprocessing: Handling missing values, encoding categorical variables, and normalizing numeric data
Deployment: Real-time predictions based on user input

Financial Logic:

Evaluates suitability of investment options based on factors like income, debt, savings, and goals

Challenges

Data Preprocessing: Handling categorical variables, missing data, and normalization for ML input
Model Overfitting: Adjusting architecture, learning rates, and regularization to generalize on unseen data
Model Loading Issues: Resolving mismatched layer sizes during model deployment
Integrating Financial Logic: Ensuring predictions align with real-world financial concepts (risk tolerance, debt-to-income ratio, savings goals)

Accomplishments

Developed an AI model that predicts personalized investment amounts accurately
Created a comprehensive synthetic dataset representing diverse financial profiles
Built a real-time interface for users to input data and get recommendations instantly
Fine-tuned model to overcome overfitting and improve prediction accuracy

What We Learned

Advanced data preprocessing techniques for financial datasets
Hands-on experience training regression models with PyTorch, TensorFlow, and scikit-learn
Applying financial logic to ML predictions
Deploying pre-trained models for real-time user recommendations