This master thesis work outlines the approach taken to solve the RecSys Challenge 2019, whose goal is to analyze the data of a user session to predict the hotel that will be clicked at the end. The first part of the thesis focuses on a deep understanding of the given problem, by means of an accurate analysis of the data to extract the most significant features for the prediction. The core of this work is the algorithmic solution, that focuses on a matrix factorization along with a gradient boosting technique. At first will be discussed the Matrix Factorization model alone, and then will be proven the effectiveness of the gradient boosting integration.