REVolutionizing PAR

We built a machine-learning system showing that COVID shifted multifamily value drivers from location to price-for-quality, scale, and class-specific competition.

Comment

Abstract

This research investigates the shifting determinants of Revenue per Available Unit (RevPAR) growth in the U.S. multifamily market across two distinct economic epochs: Pre-COVID (2015–2020) and Post-COVID (2022–2025). Utilizing property-level panel data and multi-radius trade area metrics, we developed a grid of six gradient-boosted models to quantify shifts in renter preferences. Our findings reveal a fundamental transition from a location-dominated growth regime to a competitive "value-for-money" regime, characterized by a 12x increase in the importance of class-based rent positioning and a significant "scale penalty" for institutional-sized assets in the post-pandemic era.

Built With

  • geminiapi
  • lightgbm
  • optuna
  • python
  • streamlit
Share this project:

Updates