From supervised to unsupervised learning - Structuring the core components of understanding

With the increasing public relevance of machine learning, the need for corresponding educational opportunities is also growing. While theoretical concepts for structuring learning content already exist for supervised learning, there is no comparable basis for unsupervised learning. This paper examines the extent to which the so-called model concept, a theoretical model to structure the core components of understanding of supervised learning, can be transferred to unsupervised learning. Using the example of k-means cluster analysis, it is shown that the basic structure of the model concept, the so-called facets, is largely transferable, even if individual core components of understanding need to be re-differentiated in terms of content. The results provide a theoretical basis for the development of learning objectives and teaching materials for unsupervised learning and open up further questions regarding the implementation and empirical validation of the proposed structure.