Abstract
Recently, machine learning technology has been involved successfully in our life in an extreme manner in various domains. In this paper, we investigate the machine learning concept for educational data mining systems, that focus on developing new approaches to discover meaningful knowledge from stored data. Educational data come from different resources such as academic data from students, virtual courses, e-learning log files, and so on. Predicting student marks is a challenging problem in the educational sector. We applied a hybrid feature selection algorithm with different machine learning classifiers (i.e. nearest neighbors (kNN), Convolutional Neural Network (CNN), Naive Bayes (NB) and decision trees (C4.5)) to predict the student's performance. A feature selection algorithm is used to select the most valuable features. In this paper, we applied a binary genetic algorithm as a wrapper feature selection. A benchmark dataset is used from UCI Machine Learning Repository, and the obtained results show excellent performance.