Determining Student Behaviour Using Deep Learning Methods

Knowing students’ behavior in universities starts from the coordination stage for admission to the university, especially in universities where enrolment requires preliminary exams. We find that the selection process faces several problems and difficulties, for example, many students try to circumvent the rules for admission, and some of them try to apply more than once, desirous and insisting on joining the university and the increase in the number of students wishing to join the university exhausts those in charge of the process The choice, and therefore it was necessary to design an information system to manages the coordination system to overcome all the aforementioned obstacles, depending on the national number of the student applying to the university, which makes us overcome the attempts of students to log in in more than one way, and the information system distributes students to the various examination committees according to the absorptive capacity of those committees, and the information system works 24/7. The information system that we developed is based on object-oriented programming language and distributed database (DDB) as a collection of multiple logically interrelated databases distributed over a computer network. A distributed database management system (DDBMS) is used to manage distributed databases, allowing easy access to them (Amare 2018). An Intelligent system is designed to determine the degree of difficulty of the preliminary exams to select the best-advanced students based on a feedforward neural network (FFNN). It was fed with a comprehensive dataset spanning 20 years, the system was Trained with 70% of datasets then tested with 30% and was evaluation at one of the conventional Egyptian universities. Machine learning has a specific subfield known as Deep learning: deep learning often involves tens or even hundreds of successive layers of representations. In deep learning, these layered representations are (almost always) learned via models called neural networks. An Intelligent system is designed to determine the degree of difficulty of the preliminary exams to select the best-advanced students based on deep learning. In addition to, the outcome of the system indicates that the data mining algorithm has achieved a prediction success rate of 98%.