Learning Analytics on YouTube Video Learning Content Using Clustering Method

Learning resources are one of the important things that support student learning. Learning resources could be used as an amplification outside the class to the learning material delivered by the teacher. Learning videos are often used as additional learning resources However, students may have difficulties choosing learning resources since various learning resources are available. In addition, not all of the learning resources are beneficial to the students. In addition, learning resources must be directed to ensure students learn the right material and support the student’s learning success. Therefore, there is a need to understand what video would be beneficial to student learning. This study proposes an analysis of comments in a learning video to understand the student impression of the learning material. The comments are obtained using the scrapping method. Basic preprocessing is performed such as case folding, tokenization, filtering, lemmatization and/or stemming, TF-IDF weighting, and creating cosine matrix distance. Next, two clustering methods are implemented to cluster the student comments. The clustering algorithms used in this study are k Means clustering and Agglomerative Hierarchical Clustering (AHC). The elbow method is used to determine the number of clusters. The cluster results are interpreted. Further, learning analytics on the comments such as finding frequent words, listing positive and negative terms, and wordcloud described. The cluster results are evaluated using the Silhouette coefficient. The results show that AHC with average linkage yields the highest score with $\mathbf{0. 6 2 8}$. The learning analytics shows interesting insight into student learning. However, further exploration is needed.