models("Outlier") not finding the OutlierDetectionPython models

[ablaom]

As expected:

julia> models() do m m.package_name == "OutlierDetectionPython" end
17-element Vector{NamedTuple{(:name, :package_name, :is_supervised, :abstract_type, :deep_properties, :docstring, :fit_data_scitype, :human_name, :hyperparameter_ranges, :hyperparameter_types, :hyperparameters, :implemented_methods, :inverse_transform_scitype, :is_pure_julia, :is_wrapper, :iteration_parameter, :load_path, :package_license, :package_url, :package_uuid, :predict_scitype, :prediction_type, :supports_class_weights, :supports_online, :supports_training_losses, :supports_weights, :transform_scitype, :input_scitype, :target_scitype, :output_scitype), T} where T<:Tuple}:
 (name = ABODDetector, package_name = OutlierDetectionPython, ... )
 (name = CBLOFDetector, package_name = OutlierDetectionPython, ... )
 (name = COFDetector, package_name = OutlierDetectionPython, ... )
 (name = COPODDetector, package_name = OutlierDetectionPython, ... )
 (name = HBOSDetector, package_name = OutlierDetectionPython, ... )
 (name = IForestDetector, package_name = OutlierDetectionPython, ... )
 (name = KNNDetector, package_name = OutlierDetectionPython, ... )
 (name = LMDDDetector, package_name = OutlierDetectionPython, ... )
 (name = LOCIDetector, package_name = OutlierDetectionPython, ... )
 (name = LODADetector, package_name = OutlierDetectionPython, ... )
 (name = LOFDetector, package_name = OutlierDetectionPython, ... )
 (name = MCDDetector, package_name = OutlierDetectionPython, ... )
 (name = OCSVMDetector, package_name = OutlierDetectionPython, ... )
 (name = PCADetector, package_name = OutlierDetectionPython, ... )
 (name = RODDetector, package_name = OutlierDetectionPython, ... )
 (name = SODDetector, package_name = OutlierDetectionPython, ... )
 (name = SOSDetector, package_name = OutlierDetectionPython, ... )

But where are they below?

julia> models("Outlier")
9-element Vector{NamedTuple{(:name, :package_name, :is_supervised, :abstract_type, :deep_properties, :docstring, :fit_data_scitype, :human_name, :hyperparameter_ranges, :hyperparameter_types, :hyperparameters, :implemented_methods, :inverse_transform_scitype, :is_pure_julia, :is_wrapper, :iteration_parameter, :load_path, :package_license, :package_url, :package_uuid, :predict_scitype, :prediction_type, :supports_class_weights, :supports_online, :supports_training_losses, :supports_weights, :transform_scitype, :input_scitype, :target_scitype, :output_scitype), T} where T<:Tuple}:
 (name = ABODDetector, package_name = OutlierDetectionNeighbors, ... )
 (name = AEDetector, package_name = OutlierDetectionNetworks, ... )
 (name = COFDetector, package_name = OutlierDetectionNeighbors, ... )
 (name = DNNDetector, package_name = OutlierDetectionNeighbors, ... )
 (name = DSADDetector, package_name = OutlierDetectionNetworks, ... )
 (name = ESADDetector, package_name = OutlierDetectionNetworks, ... )
 (name = KNNDetector, package_name = OutlierDetectionNeighbors, ... )
 (name = LOFDetector, package_name = OutlierDetectionNeighbors, ... )
 (name = OneClassSVM, package_name = LIBSVM, ... )

[Frank-III]

My Julia and MLJ version

Julia Version 1.7.2
MLJ v0.16.7

and this function seems work fine for me

julia> models("Outlier")
25-element Vector{NamedTuple{(:name, :package_name, :is_supervised, :abstract_type, :deep_properties, :docstring, :fit_data_scitype, :hyperparameter_ranges, :hyperparameter_types, :hyperparameters, :implemented_methods, :inverse_transform_scitype, :is_pure_julia, :is_wrapper, :iteration_parameter, :load_path, :package_license, :package_url, :package_uuid, :predict_scitype, :prediction_type, :supports_class_weights, :supports_online, :supports_training_losses, :supports_weights, :transform_scitype, :input_scitype, :target_scitype, :output_scitype)}}:
 (name = ABODDetector, package_name = OutlierDetectionNeighbors, ... )
 (name = ABODDetector, package_name = OutlierDetectionPython, ... )
 (name = AEDetector, package_name = OutlierDetectionNetworks, ... )
 (name = CBLOFDetector, package_name = OutlierDetectionPython, ... )
 (name = COFDetector, package_name = OutlierDetectionNeighbors, ... )
 (name = COFDetector, package_name = OutlierDetectionPython, ... )
 (name = COPODDetector, package_name = OutlierDetectionPython, ... )
 (name = DNNDetector, package_name = OutlierDetectionNeighbors, ... )
 (name = DSADDetector, package_name = OutlierDetectionNetworks, ... )
 (name = ESADDetector, package_name = OutlierDetectionNetworks, ... )
 (name = HBOSDetector, package_name = OutlierDetectionPython, ... )
 (name = IForestDetector, package_name = OutlierDetectionPython, ... )
 (name = KNNDetector, package_name = OutlierDetectionNeighbors, ... )
 (name = KNNDetector, package_name = OutlierDetectionPython, ... )
 (name = LMDDDetector, package_name = OutlierDetectionPython, ... )
 (name = LOCIDetector, package_name = OutlierDetectionPython, ... )
 (name = LODADetector, package_name = OutlierDetectionPython, ... )
 (name = LOFDetector, package_name = OutlierDetectionNeighbors, ... )
 (name = LOFDetector, package_name = OutlierDetectionPython, ... )
 (name = MCDDetector, package_name = OutlierDetectionPython, ... )
 (name = OCSVMDetector, package_name = OutlierDetectionPython, ... )
 (name = PCADetector, package_name = OutlierDetectionPython, ... )
 (name = RODDetector, package_name = OutlierDetectionPython, ... )
 (name = SODDetector, package_name = OutlierDetectionPython, ... )
 (name = SOSDetector, package_name = OutlierDetectionPython, ... )

[ablaom]

Right, thanks @Frank-III . If I use MLJModels v0.14.14 I get the same as you. So the cause postdates that version of MLJModels.jl.

[ablaom]

I have resolved that the new behaviour, while unexpected, is correct. For example, the doc-string for the OutlierDetectionPython model called KNNDetector does not contain the string "Outlier":

julia> v=models("KNNDetector")
2-element Vector{NamedTuple{(:name, :package_name, :is_supervised, :abstract_type, :deep_properties, :docstring, :fit_data_scitype, :human_name, :hyperparameter_ranges, :hyperparameter_types, :hyperparameters, :implemented_methods, :inverse_transform_scitype, :is_pure_julia, :is_wrapper, :iteration_parameter, :load_path, :package_license, :package_url, :package_uuid, :predict_scitype, :prediction_type, :supports_class_weights, :supports_online, :supports_training_losses, :supports_weights, :transform_scitype, :input_scitype, :target_scitype, :output_scitype), T} where T<:Tuple}:
 (name = KNNDetector, package_name = OutlierDetectionNeighbors, ... )
 (name = KNNDetector, package_name = OutlierDetectionPython, ... )

julia> knn2 = v[2];

julia> Markdown.parse(knn2.docstring)
  KNNDetector(n_neighbors = 5,
           method = "largest",
           radius = 1.0,
           algorithm = "auto",
           leaf_size = 30,
           metric = "minkowski",
           p = 2,
           metric_params = nothing,
           n_jobs = 1)

  https://pyod.readthedocs.io/en/latest/pyod.models.html#module-pyod.models.knn
  (https://pyod.readthedocs.io/en/latest/pyod.models.html#module-pyod.models.knn)

(while the other KNNDetector does). The change in the behaviour is either because of a change in the docstring (which I could not find) or quite possibly this MLJModelInterface change (last item). We had decided changes to docstrings should not be regarded as breaking.

[ablaom]

When the docstrings for OutlierDetection package are updated a more expected (and useful) behavior ought to return.