Report that clusters students who are similar within the cluster, different between clusters

[IvonArroyo]

It is time to add some differentiation. For the time being we could start by creating 2 different clusters of students (by default) using K-Means clustering.

I think Boming could create a method in MathSpring that generates a certain number of clusters (parameter) in a class, and returns the k clusters requested (k lists of students).

Then Frank can take this method and merge it into the Teacher tools as a report. Teachers should be able to specify how many clusters they want (e.g. 2, 3, or 4).

The result of the report should be a summary of how many students are in each cluster, and how they are doing along several variables: average Effort, average Mastery per topic, average emotion values.