search

ShobiStassen / PARC

MIT License
41 stars 11 forks source link

Controlling the number of clusters #9

Closed ajitjohnson closed 4 years ago

ajitjohnson commented 4 years ago

In Leiden clustering, we could pass a resolution parameter value to control the coarseness of the clustering. Is it possible to expose that parameter?

ShobiStassen commented 4 years ago

Hi Ajit, sure we can expose the resolution parameter. That's a good idea. Will try to do it today and let you know when it's done ! In case you are trying to uncover some small populations, the level of graph pruning is quite impactful. (The supplementary materials of our paper has more details on this) Shobi

ajitjohnson commented 4 years ago

Thank you @ShobiStassen I am interested in the contrary actually- to reduce the number of total clusters.

ShobiStassen commented 4 years ago

hi, i just added the resolution_parameter, do let me know if it works for you. If you are trying to reduce the number of clusters, perhaps also try to set both pruning parameters to a value of 1 or 2 (which minimizes the extent of pruning). You can also increase K (NN), the method is still very fast.

ajitjohnson commented 4 years ago

Thank you for your super quick implementation. Works perfectly. Thank you.