The surging popularity of Airbnb demands a deeper understanding of guest behavior in diverse markets. While existing research had predominantly focused on major players like the US and Europe, there was a significant gap in the study of the Asian travel landscape, particularly the sizeable Japanese Airbnb market. To bridge this knowledge gap, the researchers employed machine learning techniques to predict the Airbnb's binary overall rating status from listing-level star sub-scores and identify key drivers of guest satisfaction in Japan. By utilizing Support Vector Machines (SVM), Decision Trees (DT), Random Forests (RF), and Logistic Regression (LR) classifiers, we found that logistic regression was the most effective model, with an accuracy rate of 98.92%. The high classification accuracy was expected given the strong correlations among listing-level star sub-score categories, which shares a common measurement structure with the overall rating. Feature importance analysis indicates that Accuracy (24.39%), Value (23.80%), and Communication (16.79%) are the most influential factors in predicting guest satisfaction outcomes. This study is the first to emphasize the paramount importance of the Accuracy factor in the Asian Airbnb market. Overall, Japan garnered 98.51% positive feedback from Airbnb ratings, with Communication receiving the highest satisfaction rating at 99.06%. Airbnb hosts in Japan can utilize these insights to enhance guest experiences and support the tourism sector. • Identified guest preferences through classifiers and feature importance • Classified 14,802 Airbnb Japan reviews based on four machine learning classifiers • Logistic regression was the top classifier, attaining an accuracy of 98.80% • The crucial importance of accuracy (28%) in managing Japan's Airbnb listings • 88% of guest feedback was positive, indicating high satisfaction
Fernandez et al. (Thu,) studied this question.