[BUG]: Argument Error when running predict through MLJ interface.

[KevinG1002]

What happened?

Hello!

I was trying to run the following command MLJ.predict(mach, (data = periodic_X, idx=params.best_idx)) taken from this library's basic tutorial but ran into the following error:

ERROR: LoadError: ArgumentError: `matrix` method only supports AbstractMatrix or containers implementing the Tables interface.
Stacktrace:
  [1] _matrix(::MLJModelInterface.FullInterface, ::Val{:other}, X::NamedTuple{(:data, :idx), Tuple{Matrix{Float64}, Int64}}; kw::Base.Pairs{Symbol, Bool, Tuple{Symbol}, NamedTuple{(:transpose,), Tuple{Bool}}})
    @ MLJModelInterface ~/.julia/packages/MLJModelInterface/y9x5A/src/data_utils.jl:49
  [2] matrix(m::MLJModelInterface.FullInterface, v::Val{:other}, X::NamedTuple{(:data, :idx), Tuple{Matrix{Float64}, Int64}}; kw::Base.Pairs{Symbol, Bool, Tuple{Symbol}, NamedTuple{(:transpose,), Tuple{Bool}}})
    @ MLJModelInterface ~/.julia/packages/MLJModelInterface/y9x5A/src/data_utils.jl:45
  [3] matrix(X::NamedTuple{(:data, :idx), Tuple{Matrix{Float64}, Int64}}; kw::Base.Pairs{Symbol, Bool, Tuple{Symbol}, NamedTuple{(:transpose,), Tuple{Bool}}})
    @ MLJModelInterface ~/.julia/packages/MLJModelInterface/y9x5A/src/data_utils.jl:38
  [4] get_matrix_and_info(X::NamedTuple{(:data, :idx), Tuple{Matrix{Float64}, Int64}}, #unused#::Type{DynamicQuantities.SymbolicDimensions{DynamicQuantities.FixedRational{Int32, 25200}}})
    @ SymbolicRegression.MLJInterfaceModule ~/.julia/packages/SymbolicRegression/5HYEt/src/MLJInterface.jl:218
  [5] predict(m::SRRegressor{DynamicQuantities.SymbolicDimensions{DynamicQuantities.FixedRational{Int32, 25200}}, DataType, false}, fitresult::NamedTuple{(:state, :num_targets, :options, :variable_names, :y_variable_names, :y_is_table, :X_units, :y_units, :types), Tuple{Tuple{Vector{Vector{SymbolicRegression.PopulationModule.Population{Float64, Float64}}}, SymbolicRegression.HallOfFameModule.HallOfFame{Float64, Float64}}, Int64, SymbolicRegression.CoreModule.OptionsStructModule.Options{Int64, DynamicExpressions.OperatorEnumModule.OperatorEnum, false, Optim.Options{Float64, Nothing}, StatsBase.Weights{Float64, Float64, Vector{Float64}}}, Vector{String}, Nothing, Bool, Nothing, Nothing, NamedTuple{(:T, :X_t, :y_t, :w_t, :state, :X_units, :y_units, :X_units_clean, :y_units_clean), NTuple{9, DataType}}}}, Xnew::NamedTuple{(:data, :idx), Tuple{Matrix{Float64}, Int64}})
    @ SymbolicRegression.MLJInterfaceModule ~/.julia/packages/SymbolicRegression/5HYEt/src/MLJInterface.jl:342
  [6] predict(mach::Machine{SRRegressor{DynamicQuantities.SymbolicDimensions{DynamicQuantities.FixedRational{Int32, 25200}}, DataType, false}, SRRegressor{DynamicQuantities.SymbolicDimensions{DynamicQuantities.FixedRational{Int32, 25200}}, DataType, false}, true}, Xraw::NamedTuple{(:data, :idx), Tuple{Matrix{Float64}, Int64}})
    @ MLJBase ~/.julia/packages/MLJBase/7nGJF/src/operations.jl:135

It is worth noting that periodic_X is of type Matrix{Float64} and mach is my fitted model. Any chance you've seen this before?

Best,

Kevin

Version

0.22.2

Operating System

macOS

Interface

Script (i.e., python my_script.py), Julia REPL

Relevant log output

ERROR: LoadError: ArgumentError: `matrix` method only supports AbstractMatrix or containers implementing the Tables interface.
Stacktrace:
  [1] _matrix(::MLJModelInterface.FullInterface, ::Val{:other}, X::NamedTuple{(:data, :idx), Tuple{Matrix{Float64}, Int64}}; kw::Base.Pairs{Symbol, Bool, Tuple{Symbol}, NamedTuple{(:transpose,), Tuple{Bool}}})
    @ MLJModelInterface ~/.julia/packages/MLJModelInterface/y9x5A/src/data_utils.jl:49
  [2] matrix(m::MLJModelInterface.FullInterface, v::Val{:other}, X::NamedTuple{(:data, :idx), Tuple{Matrix{Float64}, Int64}}; kw::Base.Pairs{Symbol, Bool, Tuple{Symbol}, NamedTuple{(:transpose,), Tuple{Bool}}})
    @ MLJModelInterface ~/.julia/packages/MLJModelInterface/y9x5A/src/data_utils.jl:45
  [3] matrix(X::NamedTuple{(:data, :idx), Tuple{Matrix{Float64}, Int64}}; kw::Base.Pairs{Symbol, Bool, Tuple{Symbol}, NamedTuple{(:transpose,), Tuple{Bool}}})
    @ MLJModelInterface ~/.julia/packages/MLJModelInterface/y9x5A/src/data_utils.jl:38
  [4] get_matrix_and_info(X::NamedTuple{(:data, :idx), Tuple{Matrix{Float64}, Int64}}, #unused#::Type{DynamicQuantities.SymbolicDimensions{DynamicQuantities.FixedRational{Int32, 25200}}})
    @ SymbolicRegression.MLJInterfaceModule ~/.julia/packages/SymbolicRegression/5HYEt/src/MLJInterface.jl:218
  [5] predict(m::SRRegressor{DynamicQuantities.SymbolicDimensions{DynamicQuantities.FixedRational{Int32, 25200}}, DataType, false}, fitresult::NamedTuple{(:state, :num_targets, :options, :variable_names, :y_variable_names, :y_is_table, :X_units, :y_units, :types), Tuple{Tuple{Vector{Vector{SymbolicRegression.PopulationModule.Population{Float64, Float64}}}, SymbolicRegression.HallOfFameModule.HallOfFame{Float64, Float64}}, Int64, SymbolicRegression.CoreModule.OptionsStructModule.Options{Int64, DynamicExpressions.OperatorEnumModule.OperatorEnum, false, Optim.Options{Float64, Nothing}, StatsBase.Weights{Float64, Float64, Vector{Float64}}}, Vector{String}, Nothing, Bool, Nothing, Nothing, NamedTuple{(:T, :X_t, :y_t, :w_t, :state, :X_units, :y_units, :X_units_clean, :y_units_clean), NTuple{9, DataType}}}}, Xnew::NamedTuple{(:data, :idx), Tuple{Matrix{Float64}, Int64}})
    @ SymbolicRegression.MLJInterfaceModule ~/.julia/packages/SymbolicRegression/5HYEt/src/MLJInterface.jl:342
  [6] predict(mach::Machine{SRRegressor{DynamicQuantities.SymbolicDimensions{DynamicQuantities.FixedRational{Int32, 25200}}, DataType, false}, SRRegressor{DynamicQuantities.SymbolicDimensions{DynamicQuantities.FixedRational{Int32, 25200}}, DataType, false}, true}, Xraw::NamedTuple{(:data, :idx), Tuple{Matrix{Float64}, Int64}})
    @ MLJBase ~/.julia/packages/MLJBase/7nGJF/src/operations.jl:135

Extra Info

No response

[MilesCranmer]

That feature was not implemented until recently. Could you try the current version, v0.24.5? (You mentioned you are using v0.22.2)

[KevinG1002]

Hi @MilesCranmer -- confirming that it works :-) Thank you!