Towards a Machine Learning and Datamining approach to identify customer satisfaction factors on Airbnb

Most companies make customer satisfaction a top priority. Indeed, satisfaction is considered as one of the pillars of customer loyalty in order to bring them to use a product or a service again and the tourism sector is not an exception. In fact, all the companies working in this sector strive to provide the best possible service to their customers in order to gain their satisfaction, in this case the companies specialized in P2P accommodation rental such as Airbnb. In the existing literature there are many works that have tried to identify the identifiers of customer satisfaction. However, we have not found any study that tries to capture these identifiers by taking into account the category to which the customer belongs, namely individuals, couples and families. In our study we analyzed 100,000 reviews left by customers on the Airbnb platform towards accommodations located in London. We adopted an approach based on Natural Language Processing (NLP) to segment the reviews according to the categories of the customers, then we trained two regression models; Multiple Linear Regression and Gradient Boosting Regression to determine the weights of the 6 elementary scores noted on Airbnb. The results show that the importance given to each elemental indicator changes depending on the category the customer belongs to, and sometimes even remarkably, which can have important implications for the P2P hosting industry by refining offers according to the customer's category in order to ensure the best possible experience.