Predicting Airbnb Listing Price Across Different Cities

Airbnb has become increasingly popular among travelers for accommodation across the world. Accordingly, there are large datasets being collected from the Airbnb listings with rich features. In this project, we aim to predict Airbnb listing price in three cities – New York City (NYC), Paris and Berlin with various machine learning approaches. With one of our best approach – neural network, we have achieved r-squared values of 0.769 in train and 0.741 in test on the NYC dataset, values of 0.762 in train and 0.716 in test on the Paris dataset, and values of 0.816 in train and 0.773 in test on the combined dataset. We showed that neural network that is trained on the combined dataset of NYC and Paris, rather than the individual datasets, is more generalizable to predict the price in a different city – Berlin.