Philippe Modard comments

Results 47 comments of


                                            Philippe Modard

for the same set of data, the centroids vary for new run

Yes, that's linked to your problem. > Finding the global minimum of the k-means problem is [NP-hard](https://en.wikipedia.org/wiki/K-means_clustering#Complexity) in general.

for the same set of data, the centroids vary for new run

This is one [todo](https://github.com/Philmod/node-kmeans#to-do). I think that can be solved with different solutions: - [replicates](http://www.mathworks.com/help/stats/k-means-clustering.html#brah7f_-1): trying many random starting points and merging - adding some randomness I'm happy if you...

Perfomance with large elements ?

I haven't tried at this scale. Please share your experience.

cannot install with express 3.x

I have the same issue...

River_Stale exception

That's it? It took me 2 days to index this collection, and I should just retry?

River_Stale exception

@richardwilly98 How do I increase the oplog.rs size? @benmccann 30e6 entries

Multiple sensors

Hey Carl, That's a great question. This library only uses a single input variable, but you could transform your 7 temperatures into a single variable, depending on your goal: -...

High resolution timer

I guess that wouldn't hurt.

Can't Set Initial Values of Cluster Centers

That's a great idea, feel free to submit a PR :)

Feature Request: Make parameter for a maximum cluster size and remove number of clusters

Feel free to submit a PR :-)