search

fade2black / graphsrb

Simple directed and undirected weighted graphs and basic operations on them
MIT License
11 stars 0 forks source link

clustering algorithms (SNN, DBSCAN, OPTICS) #1

Open maia opened 5 years ago

maia commented 5 years ago

This is a great gem. Do you happen to have any implementations of clustering algorithms? Personally I'd be very interested in Shared Nearest Neighbor (SNN) Clustering, as I couldn't find any Ruby implementation yet, it's very useful for clustering high dimensional data (DBSCAN isn't, OPTICS is slightly better), and doesn't need to be provided a fixed number of clusters (as in k-means).

Thanks a lot!

fade2black commented 5 years ago

Hi @maia, I am glad you like it.

Haven't implemented SNN yet. Seems to be an interesting topics. I took in mind. I am planning to review the whole gem again and improve it. I will let you know as soon as I update it.

maia commented 5 years ago

Thanks a lot, I'm very much looking forward to SNN.

And if I may add other suggestions: add some comments to the readme and the examples subfolders, e.g. for each algorithm its goal, possible usages and methodological explanation/link. They all seem very lean and usable, but probably not everyone will intuitively know what they all aim to accomplish. Oh, and PageRank might also be a classic algorithm to add (which does have a few ruby implementations, but mostly in abandoned, outdated gems).