Short-term rental / investment analytics
Data Engineering · Machine Learning · Financial Modeling
Portugal Rental Investment Intelligence
A financial decision-support system that evaluates whether a short-term rental property in Portugal is a viable investment. It ingests multi-platform listing data (Airbnb, Booking, Vrbo), models occupancy and nightly price, and simulates revenue, cash flow, and mortgage payback across regions.
Data engineering, ML modeling, financial modeling, dashboard
Python · Jupyter · scikit-learn / XGBoost · SQL · Tableau
Investors compare short-term rental markets on intuition and scattered listings. Whether a purchase in a given Portuguese region actually pays back depends on occupancy, nightly price, expenses, and financing, variables that live across platforms and are hard to combine into one view.
The goal was a quantitative decision-support system, not a descriptive report: expected annual revenue, break-even occupancy, monthly cash flow, mortgage sustainability, years to repay a €150,000 investment, and how scaling to multiple properties changes the outcome.
Architecture
SOURCE: Airbnb · Booking · Vrbo
Public multi-platform listing snapshots for Portuguese regions
Bronze: raw ingestion
Listings, availability, coordinates, and reviews captured as received
Silver: filtered & conformed
Geographic filtering by coordinates and municipality, cleaning, cross-platform matching
Gold: warehouse & features
Validated warehouse tables and model features
ML: occupancy & nightly price
XGBoost occupancy and nightly-price models, versioned and benchmarked
Financial model & decision view
Revenue, cash flow, €150k mortgage payback, and regional comparison
- Custom ingestion pipeline for Airbnb, Booking, and Vrbo listing snapshots
- Medallion architecture (Bronze / Silver / Gold) with geographic filtering by coordinates and municipality
- Occupancy estimated from availability, (365 − availability_365) / 365, across conservative, base, and optimistic scenarios
- XGBoost models for occupancy and nightly price, versioned with benchmark metadata
- Financial modeling: revenue, expenses, cash flow, and a €150,000 mortgage simulation with interest-rate sensitivity
- Cross-market comparison across the Lisbon coast, Porto, and the Douro wine-tourism region, with a dashboard output
A repeatable investment-intelligence system that turns scattered listing data into a financial view an investor can act on, comparing regions and modeling payback instead of guessing. It spans the full path MCC works across: ingestion, data engineering, machine learning, financial modeling, and a decision-ready interface.
Got a pricing or investment question buried in your data?
Let's turn it into models and numbers you can actually act on. Want to talk it through?