promote input data type to float64

[floswald]

using GaussianMixtures, RDatasets
julia> f = dataset("datasets","faithful")
272×2 DataFrame
│ Row │ Eruptions │ Waiting │
│     │ Float64   │ Int64   │
├─────┼───────────┼─────────┤
│ 1   │ 3.6       │ 79      │
│ 2   │ 1.8       │ 54      │
│ 3   │ 3.333     │ 74      │
│ 4   │ 2.283     │ 62      │
│ 5   │ 4.533     │ 85      │
...

julia> GMM(2,f.Waiting)
[ Info: Initializing GMM, 2 Gaussians LinearAlgebra.diag covariance 1 dimensions using 272 data points
  Iters               objv        objv-change | affected 
-------------------------------------------------------------
      0       1.350600e+04
      1       8.878464e+03      -4.627536e+03 |        2
      2       8.855791e+03      -2.267287e+01 |        0
      3       8.855791e+03       0.000000e+00 |        0
K-means converged with 3 iterations (objv = 8855.79069767458)
ERROR: InexactError: Int64(54.75)
Stacktrace:
 [1] Type at ./float.jl:703 [inlined]
 [2] convert at ./number.jl:7 [inlined]
 [3] setindex! at ./array.jl:767 [inlined]
 [4] copyto!(::IndexLinear, ::Array{Int64,2}, ::IndexCartesian, ::LinearAlgebra.Adjoint{Float64,Array{Float64,2}}) at ./abstractarray.jl:764
 [5] copyto! at ./abstractarray.jl:745 [inlined]
 [6] Type at ./array.jl:482 [inlined]
 [7] convert(::Type{Array{Int64,2}}, ::LinearAlgebra.Adjoint{Float64,Array{Float64,2}}) at ./array.jl:474
 [8] #GMMk#9(::Symbol, ::Int64, ::Int64, ::Int64, ::Function, ::Int64, ::Array{Int64,2}) at /Users/florian.oswald/.julia/dev/GaussianMixtures/src/train.jl:110
 [9] (::getfield(GaussianMixtures, Symbol("#kw##GMMk")))(::NamedTuple{(:kind, :nInit, :nIter, :sparse),Tuple{Symbol,Int64,Int64,Int64}}, ::typeof(GaussianMixtures.GMMk), ::Int64, ::Array{Int64,2}) at ./none:0
 [10] #GMM#7(::Symbol, ::Symbol, ::Int64, ::Int64, ::Int64, ::Int64, ::Type, ::Int64, ::Array{Int64,2}) at /Users/florian.oswald/.julia/dev/GaussianMixtures/src/train.jl:37
 [11] (::getfield(Core, Symbol("#kw#Type")))(::NamedTuple{(:method, :kind, :nInit, :nIter, :nFinal, :sparse),Tuple{Symbol,Symbol,Int64,Int64,Int64,Int64}}, ::Type{GMM}, ::Int64, ::Array{Int64,2}) at ./none:0
 [12] #GMM#8(::Symbol, ::Int64, ::Int64, ::Int64, ::Int64, ::Type, ::Int64, ::Array{Int64,1}) at /Users/florian.oswald/.julia/dev/GaussianMixtures/src/train.jl:43
 [13] GMM(::Int64, ::Array{Int64,1}) at /Users/florian.oswald/.julia/dev/GaussianMixtures/src/train.jl:43
 [14] top-level scope at none:0

julia> y = convert(Vector{Float64},f.Waiting)

julia> GMM(2,y)
[ Info: Initializing GMM, 2 Gaussians LinearAlgebra.diag covariance 1 dimensions using 272 data points
  Iters               objv        objv-change | affected 
-------------------------------------------------------------
      0       1.790200e+04
      1       8.942126e+03      -8.959874e+03 |        2
      2       8.858336e+03      -8.378957e+01 |        2
      3       8.855791e+03      -2.545360e+00 |        0
      4       8.855791e+03       0.000000e+00 |        0
K-means converged with 4 iterations (objv = 8855.79069767458)
┌ Info: K-means with 272 data points using 4 iterations
└ 68.0 data points per parameter
[ Info: Running 10 iterations EM on diag cov GMM with 2 Gaussians in 1 dimensions
[ Info: iteration 1, average log likelihood -3.802376
[ Info: iteration 2, average log likelihood -3.801642
[ Info: iteration 3, average log likelihood -3.801544
[ Info: iteration 4, average log likelihood -3.801506
[ Info: iteration 5, average log likelihood -3.801489
[ Info: iteration 6, average log likelihood -3.801482
[ Info: iteration 7, average log likelihood -3.801479
[ Info: iteration 8, average log likelihood -3.801478
[ Info: iteration 9, average log likelihood -3.801477
[ Info: iteration 10, average log likelihood -3.801477
┌ Info: EM with 272 data points 10 iterations avll -3.801477
└ 54.4 data points per parameter
GMM{Float64} with 2 components in 1 dimensions and diag covariance
Mix 1: weight 0.639023
  mean: [80.093]
  variance: [34.4079]
Mix 2: weight 0.360977
  mean: [54.6179]
  variance: [34.5016]

[davidavdav]

did the commit help?