The Diaspora Investment Advisor is an explainable machine learning recommendation system designed to help diaspora investors identify investment opportunities that align with their financial goals and constraints.
Rather than predicting returns, the system focuses on matching investor preferences to suitable investment options across asset classes such as stocks, bonds, real estate, and funds using transparent, human-readable logic.
Diaspora investors often face challenges such as:
- Limited visibility into trustworthy investment options
- Difficulty aligning risk tolerance with available opportunities
- Unclear liquidity and lock-in constraints
- Opaque recommendation logic on existing platforms
This project demonstrates how machine learning can support better financial decision-making by producing clear, preference-aware recommendations.
The system follows a structured ML pipeline:
The user provides:
- Risk tolerance (1–5)
- Minimum and maximum investment amount
- Investment horizon (short / medium / long)
- Liquidity preference (low / medium / high)
- Preferred asset classes
These inputs form the investor profile.
Each investment option is represented with structured features:
- Asset class
- Risk level
- Minimum investment amount
- Liquidity level
- Lock-in period
This enables direct comparison between investors and investments.
Investments are filtered to remove options that:
- Fall outside the user’s budget
- Exceed the investment horizon
- Do not match selected asset classes
This ensures recommendations are feasible and relevant.
The system:
- Encodes investor preferences and investment features numerically
- Computes similarity scores using cosine similarity
- Ranks investment options from best to weakest match
After recommendations are displayed, users may request an explanation showing:
- Why each investment matches their risk tolerance
- How liquidity preferences were satisfied
- How the investment horizon was respected
- Why the option fits within their budget
This promotes trust and transparency.
- Preference-based investment recommendations
- Explainable ML decision logic
- Clear input ranges and validation
- Beginner-friendly, readable code
- Designed for easy extension and experimentation
- Python
- Pandas
- NumPy
- Scikit-learn
- Ranked investment recommendations
- Compatibility scores
- Optional plain-language explanations for each recommendation
- This project focuses on recommendation logic, not payments or execution
- It does not perform price forecasting or return prediction
- Data used is synthetic and for educational purposes
This project does not constitute financial advice.