search

JuliaAI / MLJLinearModels.jl

Generalized Linear Regressions Models (penalized regressions, robust regressions, ...)
MIT License
80 stars 13 forks source link

`predict` throws an error for `MultinomialClassifier` on crabs dataset #129

Closed ablaom closed 1 year ago

ablaom commented 1 year ago 
using MLJ
using MLJLinearModels
data = MLJ.load_crabs();
y_, X = unpack(data, ==(:sp), col->col in [:FL, :RW]);
y = coerce(y_, MLJ.OrderedFactor);
model = MultinomialClassifier()
mach = machine(model, X, y) |> fit!

julia> predict(mach, X)
ERROR: DimensionMismatch("Probability array is incompatible with the number of classes, 2, which should be equal to `4`, the last dimension of the probability array. Perhaps you meant to set `augment=true`? ")
Stacktrace:
  [1] _UnivariateFinite(support::CategoricalArrays.CategoricalVector{String, UInt32, String, CategoricalArrays.CategoricalValue{String, UInt32}, Union{}}, probs::Matrix{Float64}, N::Int64; augment::Bool, kwargs::Base.Pairs{Symbol, Union{}, Tuple{}, NamedTuple{(), Tuple{}}})
    @ CategoricalDistributions ~/.julia/packages/CategoricalDistributions/eSHdo/src/types.jl:401                                                                                
  [2] #_UnivariateFinite#17
    @ ~/.julia/packages/CategoricalDistributions/eSHdo/src/types.jl:483 [inlined]
  [3] _UnivariateFinite(::Val{true}, support::CategoricalArrays.CategoricalVector{String, UInt32, String, CategoricalArrays.CategoricalValue{String, UInt32}, Union{}}, probs::Matrix{Float64})                                                                                  
    @ CategoricalDistributions ~/.julia/packages/CategoricalDistributions/eSHdo/src/types.jl:483                                                                                
  [4] UnivariateFinite(support::CategoricalArrays.CategoricalVector{String, UInt32, String, CategoricalArrays.CategoricalValue{String, UInt32}, Union{}}, probs::Matrix{Float64}; kwargs
::Base.Pairs{Symbol, Union{}, Tuple{}, NamedTuple{(), Tuple{}}})                           
    @ CategoricalDistributions ~/.julia/packages/CategoricalDistributions/eSHdo/src/types.jl:383                                                                                
  [5] UnivariateFinite
    @ ~/.julia/packages/CategoricalDistributions/eSHdo/src/types.jl:372 [inlined]
  [6] #UnivariateFinite#25
    @ ~/MLJ/MLJBase/src/interface/data_utils.jl:137 [inlined]
  [7] UnivariateFinite
    @ ~/MLJ/MLJBase/src/interface/data_utils.jl:137 [inlined]
  [8] #UnivariateFinite#14
    @ ~/MLJ/MLJModelInterface/src/data_utils.jl:594 [inlined]
  [9] UnivariateFinite(support::CategoricalArrays.CategoricalVector{String, UInt32, String, CategoricalArrays.CategoricalValue{String, UInt32}, Union{}}, probs::Matrix{Float64})
    @ MLJModelInterface ~/MLJ/MLJModelInterface/src/data_utils.jl:594
 [10] predict(m::MultinomialClassifier, ::Tuple{Vector{Float64}, Tuple{Symbol, Symbol}, CategoricalArrays.CategoricalVector{String, UInt32, String, CategoricalArrays.CategoricalValue{String, UInt32}, Union{}}, Int64}, Xnew::NamedTuple{(:FL, :RW), Tuple{Vector{Float64}, Vector{Float64}}})                                      
    @ MLJLinearModels ~/.julia/packages/MLJLinearModels/YvwMg/src/mlj/interface.jl:92
 [11] predict(mach::Machine{MultinomialClassifier, true}, Xraw::NamedTuple{(:FL, :RW), Tuple{Vector{Float64}, Vector{Float64}}})                                      
    @ MLJBase ~/MLJ/MLJBase/src/operations.jl:133
 [12] top-level scope
    @ REPL[32]:1
 [13] top-level scope
    @ ~/.julia/packages/CUDA/DfvRa/src/initialization.jl:52