Research question: Given a listing's attributes and the timing of the stay, can a model recommend a nightly price that captures demand-driven pricing opportunities a host would otherwise miss?
Hosts on short-term rental platforms typically set one static price, or accept the platform's own suggestion without adjustment — leaving money on the table on high-demand nights and sitting empty on slow ones. This project builds a regression model that recommends a price from listing quality, seasonality, and local demand signals, tested against 20% held-out data it never saw during training.
Most hosts either guess their nightly rate or take Airbnb's own suggestion blindly, leaving money on the table on high-demand nights and sitting empty on slow ones.
A gradient-boosted regression model trained on a synthetic dataset — 400 listings across 8 Hyderabad neighborhoods, 90 days of calendar data each. Real-data upgrade path: Inside Airbnb's free public export.