Allow features to be AbstractVector (in addition to AbstractMatrix)#150
Allow features to be AbstractVector (in addition to AbstractMatrix)#150sanderbboisen wants to merge 9 commits intoJuliaAI:devfrom
AbstractVector (in addition to AbstractMatrix)#150Conversation
AbstractVector (in addition to AbstractMatrix)
|
I don't see any harm in this enhancement, apart from a mild complication in the codebase. Anyone else have objections? @sanderbboisen Would you be willing to add a test of native and ScikitLearn API with a vector |
Sure, I will get on it. Incoming update. |
Codecov Report
@@ Coverage Diff @@
## dev #150 +/- ##
======================================
Coverage ? 88.84%
======================================
Files ? 9
Lines ? 986
Branches ? 0
======================================
Hits ? 876
Misses ? 110
Partials ? 0 Continue to review full report at Codecov.
|
…e.jl into VecOrMat_support
|
I ran into an issue as several functions like I have added tests to tests/scikitlearn.jl and everything seems to work. I am sorry for the confusing commit structure, things got a little messy at some point. I think there is a general issue with the way that the module is constructed, and I have been refactoring it in a separate branch on my fork. For one, the structure does not allow multiple usage of utils.jl and there is generally some issues with code repetition iirc. I will open another PR on this when I am done refactoring. |
|
I am happy with this now. I suspect this smaller PR may create a few merge conflicts for the more complicated #166, so let's wait for: before merging this one. @sanderbboisen Thanks for your contribution and patience. |
|
@sanderbboisen Would you now be happy to rebase your PR? |
I was playing around with DecisionTree.jl for a potential project and found that I could not produce a regression tree from a single feature. I have made changes to allow
featuresto be a vector in thebuild_treefunctions.I have tested the changes using RunTest.jl with Julia 1.6.4 .