This animation shows the first 7 iterations of Chow-Liu Recursive Grouping (CLRG) latent tree learning method and the final result. The algorithm starts from the Minimal Spanning Tree (MST) and adds hidden nodes while iterating over the non-leaf nodes.
The nodes in the tree are knowledge components covered in an online psychology course on coursera. There are a total number of 226 knowledge components in the tree The latent tree itself is conditioned on a set of relevant covariates. To watch the latent tree grow, press the play botton.
The blue nodes correspond to the knowledge components and their labels indicate each individual knowledge component. The yellow nodes are the latent (hidden) variables labeled with "h". You can zoom into the tree for a closer look at the concepts and how they are grouped.