More and more Internet of Things (IoT) is present in human being diary life and their belongings. Because of this evolution, large quantities of data of different types are generated.Due to the IoT has as premise the virtual modeling of things in physical world, these data can be treated and used for different purposes. When Recommender Systems used data originating from IoT to characterize the context where are inserted, the concept of Recommender Systems Context-awareness arise. Although a great part of systems context-awareness use these data for increase the quality of suggestions, context data can also determinate situations where the system must make a proactive recommendation, in other words, the system provides a recommendation without explicit request of user. Once the Recommender System has autonomy to provide suggestions, the concept of Proactive Recommender System arise. Nowadays few researches applied the aspect of proactivity to provide recommendations, thus there are large gaps about the characteristics that must be considerate when develop these systems and how these characteristics are interrelated. The main problem addressed in this dissertation it is the identification of which difficulties and aspects inherent in the development of Proactive Recommender Systems. For this purpose, articles involving studies about Internet of Things, Recommender Systems, Recommender Systems Context-awareness, Proactive Recommender Systems and Machine Learning methods, are debated and serve as theoretical basis of this research. Once the principal aspects and difficulties are identified, was developed an evaluation model called Proactive Recommender System Context-Awareness, where this model is divide into four parts. For application of this evaluation model, a ProactiveRecommender System of movies, that use simulated IoT data and Inductive Machine Learning, was elaborated. As a result, the evaluation model was deployed with the principal aspects that characterize the Proactive Recommender System. Besides that, was developed a Proactive Recommender System Context-Awareness of movies. The algorithm C4.5 was used for decision trees generation that provide users' preferences profile of movies and to determinate propitious moments for proactive recommendations of movies.