In this thesis I examine how Artificial Intelligence (AI) techniques can help Computer Science students learn programming and mathematics skills more efficiently using algorithms and concepts such as Predictive Modelling, Machine Learning, Deep Learning, Representational Learning, Recommender Systems and Graph Theory. For that, I use Learning Analytics (LA) and Educational Data Mining (EDM) principles. In Learning Analytics one collects and analyses data about students and their contexts for purposes of understanding and improving their learning and the environments students interact with. Educational Data Mining applies Data Mining, Machine Learning and statistics to data captured during these learning processes. My central research question is how we can optimise the learning by students, of subjects like computer programming and mathematics in blended and online classrooms by mining and analysing data generated in these environments by the students. To validate the research question I have implemented several examples of monitoring student behaviour while learning, I have gathered various forms of student interaction data and combined it with demographics and student performance data (e.g. exam results) in order to test out different predictive models developed using a variety of AI and machine learning techniques. In these example environments I have used these models not only to predict outcome and exam performance but also to automatically generate feedback to students in a variety of ways, including recommending better programming techniques. My research question is explored by examining the performance of the AI techniques in helping to improve student learning.