Massive Open Online Courses (MOOCs) have changed the way people study by providing accessible, scalable, and convenient means of studying. Their effectiveness, nonetheless, has been marred by concerns over dropout rate and information overload. In this work, these concerns are addressed by creating personalized course recommendations through collaborative filtering strategies, such as Alternating Least Squares (ALS) and k-Nearest Neighbors (k-NN). Using the data from the KDD Cup 2015, consisting of dense learner interaction data from the XuetangX MOOC platform, these models’ performance is gauged by means of precision, recall, and F1-score. Through this work, it has been established that recommendation accuracy by means of ALS outperforms that by means of k-NN, indicating it has the ability to uncover underlying patterns from course-user interaction data that are sparse. The work indicates the ability collaborative filtering has to enhance increased learner participation, dropout rate decrease, and course completion rate boost in MOOCs. The limitations of these models, such as data sparsity, also referred to as the cold start problem, are observed, as well as future work, such as hybrid recommendation models, as well as newer approaches such as Neural Collaborative Filtering (NCF) and Graph Neural Networks (GNN). The work adds to the developing field of educational data mining, while also providing actionable recommendations toward personalization opportunities’ improvement in MOOCs.