Error saving ScikitLearn.jl pipeline

General Usage
1

Hi,

I’m trying to save a ScikitLearn pipeline but JLD is encountering a fatal error. A MWE is below. My versions are ScikitLearn@v0.7.0, JLD@v0.10.0, PyCallJLD@v0.2.1. Pkg.status(outdated=true) returns that PyCallJLD is prohibiting an update in JLD. Any advice for how I can fix the issue? Thanks in advance! (Apologies if the markdown doesn’t come through. I wasn’t clear how to format the code in the discourse.)

using ScikitLearn,JLD,PyCallJLD;

@sk_import preprocessing: (OneHotEncoder,FunctionTransformer);
@sk_import compose: ColumnTransformer;

function colshndl(X0::AbstractArray)
    ids = [2,3];   
    X = hcat(X0,X0[:,ids]);
    
    return X
end;

cols_ft = FunctionTransformer(colshndl);

ohot_ct = ColumnTransformer([("cat1",OneHotEncoder(),[0]),
                            ("cat2",OneHotEncoder(categories=[[-1]],handle_unknown="ignore"),[15]),
                            ("cat3",OneHotEncoder(categories=[[-1]],handle_unknown="ignore"),[16]),
                            ("cat4",OneHotEncoder(),[3]),
                            ("cat5",OneHotEncoder(),[4]),
                            ("cat6",OneHotEncoder(),[5]),
                            ("cat7",OneHotEncoder(categories=[[-1]],handle_unknown="ignore"),[8]),
                            ("cat8",OneHotEncoder(categories=[[-1]],handle_unknown="ignore"),[9]),
                            ("cat9",OneHotEncoder(),[13]),
                            ("cat10",OneHotEncoder(),[14])],
                            remainder="passthrough");

prepipe = Pipelines.Pipeline([("cols",cols_ft),
                              ("ohot",ohot_ct)]);
X = rand(141,15);
fit!(prepipe,X);

JLD.save("error.jld","prepipe",prepipe);

The error returned is

┌ Warning: JLD incorrectly extends FileIO functions (see FileIO documentation)
└ @ FileIO C:\Users\Colin.julia\packages\FileIO\BE7iZ\src\loadsave.jl:217
Error encountered while save FileIO.File{FileIO.DataFormat{:JLD}, String}(“error.jld”).

Fatal error:
type DataType has no field mutable