In this paper, we introduce a model that incorporates features of the fully transparent hotel booking systems and enables estimates of hotel choice probabilities in a group based on the room charges. Firstly, we extract necessary information for the estimation from big data of online booking for major four hotels near Kyoto station. Then, we consider a nested logit model as well as a multinomial logit model for the choice behavior of the customers, where the number of rooms available for booking for each hotel are possibly limited. In addition, we apply the model to an optimal room charge problem for a hotel that aims to maximize its expected sales of a certain room type in the transparent online booking systems. We show numerical examples of the maximization problem using the data of the four hotels of November 2012 which is a high season in Kyoto city. This model is useful in that hotel managers as well as hotel investors, such as hotel REITs and hotel funds, are able to predict the potential sales increase of hotels from online booking data and make use of the result as a tool for investment decisions.