Popular social networking platforms, such as Yelp, whose users can post their own reviews and follow other users' reviews of local businesses, have now become part of our daily lives. The usefulness of online reviews has led to the increase of online frauds and the analysis of their reliability is an urgent. In this paper, we classify online reviews as fake or authentic using Machine Learning Algorithms and specifically Decision Trees, Random Forests, SVMs and Neural Networks. In order to achieve this, we conducted experiments with two different datasets. The first dataset includes information for both users and reviews while the second is much larger than the first but incomplete in terms of information for both users and reviews, as we have no information about them. Thus, we compared the datasets regarding the identification of fake reviews using machine learning algorithms and we analyzed the features of users and reviews that are more important for detecting fake reviews. The results are quite encouraging and show that the users’ behavioral features, such as the number of reviews they make or the number of friends they have, play an important role in fake review detection.