Repository navigation
K-means clustering cannot currently be used in experiments #205
Description
K-means clustering is already available as a learner in ML.NET, but it is not exposed in the APIs and cannot be used in LearningPipeline.
The entry point needs to be added to the C# APIs (along with other necessary components like the clustering evaluator).
Activity
[-]Enable K-means clustering to be used in experiments[/-][+]K-means clustering cannot currently be used in experiments[/+]codemzs commentedon May 23, 2018
The entry point is already in the API. @yaeldekel has a PR out that references kmeans dll in test project do that code gen picks it up. The issue here is, I think, clusterer scorer is returning an empty IDataView from the limited debugging that I did last week but I could be wrong.
Ivanidzo4ka commentedon May 25, 2018
It works, and we have example for that. So I'm closing this issue.
accept label from user input and provide in generated code (dotnet#205)
code migration from machinelearning-automl to machinelearning repo fe…
accept label from user input and provide in generated code (dotnet#205)
Added onnx export functionality for LpNormNormalizingTransformer (#4161)
Added Onnx export functionality to PCATransformer (#4188)