Stay Elsewhere? Improving Local Search for Hotels Using Econometric Modeling and Image Classification

One of the common Web searches that have a strong local component is the search for hotel accommodation.  Customers try to identify hotels that satisfy particular criteria, such as service, food quality, and so on.  Unfortunately, today, the travel search engines provide only rudimentary ranking facilities, typically using a single ranking criterion such as distance from city center, number of stars, price per night, or, more recently, customer reviews.  This approach has obvious shortcomings. First, it ignores the multidimensional preferences of the consumer and, second, it largely ignores characteristics related to the location of the hotel, for instance, proximity to the beach or proximity to a downtown shopping area.  These location-based features represent important characteristics that influence the desirability of a particular hotel.  However, currently there are no established metrics that can isolate the importance of the location characteristics of hotels. In our work, we use the fact that the overall desirability of the hotel is reflected in the price of the rooms; therefore, using hedonic regressions, an established technique from econometrics, we estimate the weight that consumers place on different hotel characteristics.  Furthermore, since some location-based characteristics, such as proximity to the beach, are not directly measurable, we use image classification techniques to infer such features from the satellite images of the area.  Our technique is validated on a unique panel dataset consisting of 9463 different hotels located in the United States, observed over a period of 5 months.  The final outcome of our analysis allows us to compute the “residual value” of a hotel, which roughly corresponds to the “value for the money” of a particular hotel.  By ranking the hotels as using our “value for the money” approach we generate rankings that are significantly superior to existing techniques.  Our preliminary user studies show that users overwhelmingly favor the hotel rankings generated by our system.