Understanding Tourism Accommodation Performance Under Data Sparsity and High Dimensionality

The accommodation sharing market is now huge, but it is diverse and there is a wide difference in the revenue performance of properties. Understanding the features that contribute to the revenue performance of accommodation sharing is important for property owners and accommodation platforms alike. This research examines the extent to which amenities of properties contribute to revenue performance using the lens of signaling theory. Due to the large number of amenities and a sparse data set, the research employs an advanced method for variable selection, Adaptive Best Subset Solution (ABESS). The research uses a random sample of 10% of all properties from Airbnb in New Zealand. The results suggest that 28.6% of the variance in performance can be explained by 16 amenities noted on property listings (both positive and negative). The results have implications for property owners and accommodation sharing platforms in helping to improve revenue capture from listings. The research also contributes a research process for similar forms of complex data analysis.