ppalmes / combineml.jl Goto Github PK

The tag name "0.9.3" is not of the appropriate SemVer form (vX.Y.Z).
cc: @ppalmes

OneHotEncoder function has some warnings to be addressed.

When going through the notebook, there were several issues, some I could fix, some I could not.

1. use of pipeline is now prohibited (clashes with Base)
2. the RandomForest should have :num_subfeatures => 0 and not => nothing afaiu
3. I could not run the Scikitlearn cell note that both *_AVAILABLE bools are false (maybe I didn't do something but couldn't find different way)

julia> CombineML.System.LIB_SKL_AVAILABLE
false
julia> CombineML.System.LIB_CRT_AVAILABLE
false

and so SKLLearner was not available.

4. use of mean must now be accompanied by using Statistics
5. I could not run @parallel, I'm assuming it comes from using Distributed but may have changed name

The fixes on the notebook: https://github.com/tlienart/CombineML.jl , now in a PR fixing (1, 2, 4) #19

PyCall 1.90.0 is now released, which change o[:foo] and o["foo"] to o.foo and o."foo", respectively, for python objects o; see also JuliaPy/PyCall.jl#629.

The old getindex methods still work but are deprecated, so you'll want to put out a new release that uses the new methods and REQUIREs PyCall 1.90.0 to avoid having zillions of deprecation messages.

Currently, caret relies on rpy which is a wrapper in python for R. The existing implementation uses PyCall to import rpy objects which in turn being used to load caret package. By removing this dependency, caret access using RCall will be easier to maintain.

The REQUIRE file could not be found.
cc: @ppalmes

First off, thank you for maintaining this package, awesome work!!

There has been some fruitful effort recently in improving the performance (in execution time and memory allocation) for the DecisionTree package, for both classification and regression, and we're now seeing 4-10x speed up. This has required a rewrite of the _split() functions, where now the splits and hence predictions aren't identical to those of DT v0.6.5.

As a result, one of the CombineML's tests ("Pipeline works with fixture data.") is now failing, and I'm having trouble isolating the issue.

Need your help and guidance in troubleshooting this failing test and getting CombineML onto the soon to be DT v0.7.2

To reproduce the test error, just pull from master: Pkg.checkout("DecisionTree")

DimensionalityReduction.jl is deprecated and replaced by MultivariateStats.

The tag name "v.1.0.1" is not of the appropriate SemVer form (vX.Y.Z).
cc: @ppalmes

The tag name "v0.2-alpha" is not of the appropriate SemVer form (vX.Y.Z).
cc: @ppalmes

Cannot tag a new version "v0.9.3" preceding all existing versions.
cc: @ppalmes

The REQUIRE file could not be found.
cc: @ppalmes

Currently, caret wrapper is not working.

Currently, due to issues in deprecated packages, not all models are included in the test.

@JuliaRegistrator register()

FactCheck.jl is no longer maintained. It's better to just use the @test function instead of @fact.