Unsupervised machine learning approaches include factor analysis, principal component analysis, self-organising maps or clustering methods like k-means or topic models. Recall that these methods can be used both for quantitative data sets and qualitative data sets.
A common problem in machine learning is grouping the data. Factor analysis is an example of grouping variables together: The process and outcomes focus on variables. A similar process could be used to group observations-a set of variables. For example, such a process could be used to group survey responders refs or identify which new stories have similar frames (Burscher et al., 2016). Burscher et al. (2016) uses a -means clustering approach, where potential clustering positions are examined through an iterative process. On each iteration, the clusters are moved towards the mean position of all close-by observations (see Code Example 3.2).