Hi all, this could go either in statistics, data, or machine learning domains, please share. Alright, so I was trying to test various different clustering algorithms to see which one best clusters my data. In doing so, I encountered once again the odd/weird/messy interface that is currently in Clu…

thanks for letting me know, but this other post is too long to parse in detail :frowning: I tried to search for comments containing “clustering” but I didn’t find something specific relatable, only general comments about the larger ecosystem. Would you mind linking specific comments you think are no…

I’d add a field about score / probability. Many clustering algorithms are fuzzy, soft. Additional point, some algorithms maps a point to an outlier. So the label can not be limited to 1:n.

[image] Datseris: The input data can be specified two ways: <…> What about clustering arbitrary objects, when one can compute distances between them?

An alternative API is the scikit-learn like API we employ in the clustering algorithms of BetaML : model = ClusrerAlgo(options) fit!(model, data) classes = predict(model, [optional data]) The optional data is for algorithms that can generalize to new data. For example, with new data you may want …

[image] RoyiAvital: I’d add a field about score / probability. Many clustering algorithms are fuzzy, soft. The returned results type may have arbitrary algorithm-specific fields that the algorithms describe, so this is covered. if we believe there is a universal quantity that should be API -…

[image] sylvaticus: An alternative API is the scikit-learn like API we employ in the clustering algorithms of BetaML : model = ClusrerAlgo(options) fit!(model, data) classes = predict(model, [optional data]) The optional data is for algorithms that can generalize to new data. For example, wit…

[image] Datseris: result = cluster(algorithm, data) Thanks for organizing the house. Would you mind swapping the order of arguments to read as cluster(data, algo)? We adopted this convention in our stack, and it is spread across multiple packages in the ecosystem. As a non-native English spea…

[image] Datseris: The second is a heavily improved version of DBSCAN that we use in Attractors.jl but haven’t yet had the chance to put it in Clustering.jl , Man, I’d love to see a julian HDBSCAN…

[image] juliohm: Thanks for organizing the house. Would you mind swapping the order of arguments to read as cluster(data, algo)? I don’t mind at all. I actually thought that this was the language-wide convention: first have algorithm-deciding argument, then input data. This is what we have ad…

RFC: ClusteringAPI.jl

Specific Domains Statistics

jbrea March 28, 2024, 11:07pm 2

Great! Thanks for the initiative! FYI, there has also been some discussion about clustering in this long discussion on ML API.

Topic		Replies	Views
[ANN] LearnAPI.jl - Proposal for a basement-level machine learning API Package Announcements	112	7116	February 19, 2025
Package for clustering data points Machine Learning question , clustering	8	772	June 23, 2022
[ANN] BetaML.jl.. yet an other (simple) Machine Learning Package Package Announcements package , announcement , machine-learning	16	3456	May 15, 2024
Taking Fitting Seriously Data plotting	39	6013	December 8, 2018
Optimization tips for my julia code. Can I make it even faster and/or memory efficient? Performance question , python	24	4388	February 15, 2020

RFC: ClusteringAPI.jl

Related topics