[FEAT] Cluster evaluation - with ground truth data

[OlivierBinette]

Is your proposal related to a problem?

I have "ground truth" labels for a subset of my data. Specifically I have a set of clusters that I know to be correct, with nothing that should be added or removed from them.

I'd like to use these resolved clusters to evaluate the quality of my clustering results.

Describe the solution you'd like

I'd like to:

• See how my predicted clustering compares to the ground truth clustering.
• Compute precision, recall, and cluster metrics based on this set of resolved clusters.
• Compare summary statistics (e.g. the average cluster size) between my predicted clustering and the ground truth data.

Describe alternatives you've considered

Some of it is implemented in the er-evaluation package, but it's quite slow with its Pandas implementation. Some of the methods are described in my paper https://arxiv.org/pdf/2404.05622, but I think that stuff needs to be simplified.