BroadVail RevPAR Growth Prediction

Inspiration

The hotel and real estate industry underwent a structural shift during COVID, and RevPAR data clearly shows different patterns before and after this break. We wanted to both understand what changed in renter and traveler preferences and build a forward-looking tool that could support investment and development decisions.

What it does

Our project analyzes and predicts RevPAR growth across pre-COVID (2015–2020) and post-COVID (2022–2025) periods using geospatial, demographic, and trade-area features. It combines interpretable visual analysis with machine-learning predictions across multiple drive-time radii (10/15/30 minutes) for thousands of properties.

How we built it

We first created a fair “outperformance” metric by adjusting RevPAR growth relative to each market, radius, and period, allowing apples-to-apples comparisons. We then trained and evaluated multiple ML models (Random Forest, XGBoost, CatBoost), selecting CatBoost for its strong performance with categorical features, and applied it to generate growth predictions.

Challenges we ran into

Raw RevPAR growth is dominated by market-wide effects, so naive comparisons can easily lead to misleading conclusions without careful adjustment. We also had to clean and align inconsistent pre- and post-COVID data while ensuring predictions stayed consistent across multiple drive-time definitions.

Accomplishments that we're proud of

We built an end-to-end pipeline that moves from raw data to fair comparisons, interpretable insights, and scalable predictions across thousands of properties. We also demonstrated clear structural differences between pre- and post-COVID drivers while avoiding over-interpretation of small effects.

What we learned

COVID changed the relative importance of many neighborhood and property features, but most effects are subtle once market context is removed. Outperformance is generally multi-factor, with amenity density, supply pressure, building size, and location context all contributing modestly rather than one variable dominating.

What's next for BroadVail RevPAR Growth Prediction

Next, we plan to incorporate temporal dynamics, macroeconomic indicators, and tourism trends to strengthen predictive signal. Longer term, we aim to turn this into a valuation and underwriting tool for scenario testing, market selection, and investment strategy.