The recently proposed EMDE (Efficient Manifold Density Estimator) modelachieves state of-the-art results in session-based recommendation. In this workwe explore its application to Booking.com Data Challenge competition. The aimof the challenge is to make the best recommendation for the next destination ofa user trip, based on dataset with millions of real anonymized accommodationreservations. We achieve 2nd place in this competition. First, we use Cleora -our graph embedding method - to represent cities as a directed graph and learntheir vector representation. Next, we apply EMDE to predict the next userdestination based on previously visited cities and some features associatedwith each trip. We release the source code at:https://github.com/Synerise/booking-challenge.