Skip to content

Catch-all error can be quite misleading #379

@odunbar

Description

@odunbar

issue

This error message:

function throw_define_mlt()
throw(ErrorException("Unknown MachineLearningTool defined, please use a known implementation"))
end
function build_models!(mlt, iopairs, input_structure_mats, output_structure_mats, mlt_kwargs...)
throw_define_mlt()
end
function optimize_hyperparameters!(mlt)
throw_define_mlt()
end
function predict(mlt, new_inputs; mlt_kwargs...)
throw_define_mlt()
end

designed to catch unknown or unimplemented machine learnign tools can be misleading. Primarily, if one gets the data-type incorrect for the inputs in say predict(mlt, inputs...) it throws the error implying that the mlt argument was incorrect!

Should be corrected to a more specific message, and not thrown in these cases to avoid confusion.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions