Now let’s talk

about Rand Index, or RI. For example, X1 indicate 1 cluster from your

algorithm, and Y1 indicate a cluster from the ground truth. This two points, P1 and P2, belong to the

same cluster in X because they belong to X1, and the same set of points, P1 P2, also belong

to the same cluster in Y. This 2 point, P1 and P2, belong to different

clusters in X and Y. The Rand Index equals to a plus b divided

by the number of all possible pairs, which is n times n minus 1 divided by 2. If it’s 0, it means there’s no agreement

between this two clustering assignments, and if Rand Index equal to 1, that means a perfect

matching. First, we compute the distance matrix between

all pairs of points, then we initialize each point as a cluster. Then we check how many clusters are left. If 1 left, then we’re done. Otherwise, we merge 2 closest cluster into

1 cluster. Then we update the distance matrix with the

remaining clusters, then iterate and until the number of clusters become 1.

## Be First to Comment