[DOC] simple examples in documentation are broken

[rolling-robot]

Simple examples on https://outlierdetectionjl.github.io/OutlierDetection.jl/dev/API/detectors/#lofdetector
https://juliapackages.com/p/outlierdetection
do not seem to work.

How is it possible to fix them?

[rolling-robot]

looks like the culprit is in https://github.com/OutlierDetectionJL/OutlierDetectionInterface.jl/blob/68dce35f06ac4bd769e01da46a4b1e6fb50a42c7/src/base.jl#L72, where outdated interface is coded in

[rolling-robot]

I see that MLJ pipelines do not work for OutlierDetection now: #31, but they are mentioned on the title page https://juliapackages.com/p/outlierdetection (and do not work indeed) in MLJ section.

Is it appropriate to change example on title page to something like #31 (comment) ?

[davnn]

You are right that pipelines do not currently work since they were changed in MLJ. It would be an option to add learning networks to the README, but maybe its enough to show a simple MLJ-based model in the readme.

Edit: Thank you for your pull request!

[davnn]

@all-contributors add @rolling-robot for docs.

[allcontributors]

@davnn

I've put up a pull request to add @rolling-robot! 🎉

[rolling-robot]

I think that something like that would fit title page (at least it works)
But I can't find the source for it https://juliapackages.com/p/outlierdetection to correct.

using MLJ
using OutlierDetection
using OutlierDetectionData: ODDS

X, y = ODDS.load("annthyroid")

LOF = @iload LOFDetector pkg=OutlierDetectionNeighbors verbosity=0

lof = ProbabilisticDetector(LOF())
cv = StratifiedCV(nfolds=5, shuffle=true, rng=0)
r = range(lof, :(detector.k), values=[1,2,3,4,5:5:100...])

t = TunedModel(model=lof, resampling=cv, tuning=Grid(), range=r, acceleration=CPUThreads())
m = machine(t, X, y) |> fit!
report(m).best_history_entry
b = report(m).best_model
evaluate(b, X, y, resampling=cv)
ŷ=predict(m, X)

[rolling-robot]

Also this warning should be corrected, as current version is 0.3.3, according to github

🚧 Experimental, install from master branch until 0.2 is released and expect breaking changes 🚧

[davnn]

Do you, by any chance, know how to update that page?

[rolling-robot]

Well, I have found where it comes from. The page was statically generated long ago from a readme in this repo. This statically generated page is now at https://github.com/djsegal/JuliaPackages.jl/blob/2862c325ecf0c7e3da10e3cc7a210750a39914bf/data/readme_html/outlierdetection.txt

[rolling-robot]

Looks like title page for every single package on juliapackages.com is a 2 year old readme.

[rolling-robot]

I think this issue can be closed as the main problem for my OP is upstream.