Retrieve confidence/percentage of participant group membership

[Simon-Dirks]

Hi there! I noticed that with the /api/v3/math/pca2 call in client-report I get back group-clusters, containing information about the participant IDs of each member of each group. I'm wondering, however, is there also data available on the extent to which each member is part of each group.

I'm unsure exactly how the math worker currently defines a participant to be a member of a group. Does it calculate percentages (e.g., participant is 1% group A, 2% group B and 90% group C), and then apply a threshold?

One idea I had is to calculate the euclidean distance of a participant in the graph view to the centers of all other clusters, but I'm not sure if that math would be correct to define group membership confidence..

Any pointers would be greatly appreciated!