CompatHelper: bump compat for Documenter to 1 for package docs, (keep existing compat)

docs/Project.toml

@@ -6,6 +6,6 @@ MLBase = "f0e99cf1-93fa-52ec-9ecc-5026115318e0"
 
 [compat]
 DataFrames = "1"
-Documenter = "0.24"
+Documenter = "0.24, 1"
 Lasso = "0.7"
 MLBase = "0.9"

docs/make.jl

@@ -11,6 +11,8 @@ makedocs(
     sitename = "Lasso.jl",
     authors = "Simon Kornblith, Asaf Manela, and contributors.",
     linkcheck = !("skiplinks" in ARGS),
+    warnonly = [:missing_docs],
+    doctestfilters = [r"(?<=\.\d{5})\d+"],
     pages = [
         "Home" => "index.md",
         "Lasso paths" => "lasso.md",

docs/src/index.md

@@ -33,7 +33,7 @@ Journal of Computational and Graphical Statistics, 26:3, 525-536.
 julia> using DataFrames, Lasso
 
 julia> data = DataFrame(X=[1,2,3], Y=[2,4,7])
-3×2 DataFrames.DataFrame
+3×2 DataFrame
  Row │ X      Y
      │ Int64  Int64
 ─────┼──────────────
@@ -60,20 +60,20 @@ Coefficients:
 ────────────
     Estimate
 ────────────
-x1  3.88915 
+x1  3.88915
 x2  0.222093
 ────────────
 
 julia> predict(m)
-3-element Array{Float64,1}:
- 4.111240223622052
+3-element Vector{Float64}:
+ 4.111240223622057
  4.333333333333333
- 4.555426443044614
+ 4.555426443044609
 
 julia> predict(m, data[2:end,:])
-2-element Array{Union{Missing, Float64},1}:
+2-element Vector{Union{Missing, Float64}}:
  4.333333333333333
- 4.555426443044614
+ 4.555426443044609
 ```
 
 In the variant above, it automatically picks the size of penalty to apply to ``x_2``.

docs/src/lasso.md

@@ -71,32 +71,30 @@ step as follows:
 ```jldoctest
 julia> using DataFrames, Lasso, MLBase, Random
 
-julia> Random.seed!(124); # because CV folds are random
-
 julia> data = DataFrame(X=[1,2,3], Y=[2,4,7])
-3×2 DataFrames.DataFrame
+3×2 DataFrame
  Row │ X      Y
      │ Int64  Int64
 ─────┼──────────────
    1 │     1      2
    2 │     2      4
    3 │     3      7
 
-julia> m = fit(LassoModel, @formula(Y ~ X), data; select=MinCVmse(Kfold(3,2)))
-LassoModel using MinCVmse(Kfold([3, 1, 2], 2, 1.5)) segment of the regularization path.
+julia> m = fit(LassoModel, @formula(Y ~ X), data; select=MinCVmse(Kfold(MersenneTwister(124), 3, 2)))
+LassoModel using MinCVmse(Kfold([1, 2, 3], 2, 1.5)) segment of the regularization path.
 
 Coefficients:
 ────────────
     Estimate
 ────────────
-x1   4.33333
-x2   0.0    
+x1  -0.63377
+x2   2.48355
 ────────────
 
 julia> coef(m)
-2-element Array{Float64,1}:
- 4.333333333333333
- 0.0              
+2-element Vector{Float64}:
+ -0.6337700054337887
+  2.483551669383561
 
 ```
 

src/segselect.jl

@@ -328,8 +328,13 @@ end
 function Base.show(io::IO, obj::RegularizedModel)
     # prefix = isa(obj.m, GeneralizedLinearModel) ? string(typeof(distfun(path)).name.name, " ") : ""
     println(io, "$(typeof(obj).name.name) using $(obj.select) segment of the regularization path.")
+    println(io, "\nCoefficients:")
+    show(io, MIME("text/plain"), coeftable(obj))
+    println(io)
+end
 
-    println(io, "\nCoefficients:\n", coeftable(obj))
+function Base.show(io::IO, ::MIME"text/plain", obj::StatsModels.TableRegressionModel{<:RegularizedModel})
+    show(io, obj.model)
 end
 
 StatsBase.vcov(obj::RegularizedModel, args...) = error("variance-covariance matrix for a regularized model is not yet implemented")

test/cross_validation.jl

@@ -1,4 +1,4 @@
-using MLBase, Random
+using MLBase, Random, StableRNGs
 
 @testset "cross validation" begin
 
@@ -27,25 +27,23 @@ segminAIC = Lasso.minAIC(path)
 @test segminAIC == 72
 @test coefsAIC == β[:,segminAIC]
 
-Random.seed!(13)
-gen = Kfold(length(y),10)
+gen = Kfold(StableRNG(13), length(y), 10)
 segCVmin = cross_validate_path(path, MinCVmse(gen))
-coefsCVmin = coef(path, MinCVmse(path))
+coefsCVmin = coef(path, MinCVmse(Kfold(StableRNG(13), length(y), 10)))
 @test segCVmin == 72
 @test coefsCVmin == β[:,segCVmin]
 
-Random.seed!(13)
-gen = Kfold(length(y),10)
-segCVmin = cross_validate_path(path,X,y, MinCVmse(gen), offset=offset)
-coefsCVmin = coef(path, MinCVmse(path))
+gen = Kfold(StableRNG(13), length(y), 10)
+segCVmin = cross_validate_path(path, X, y, MinCVmse(gen), offset=offset)
+coefsCVmin = coef(path, MinCVmse(Kfold(StableRNG(13), length(y), 10)))
 @test segCVmin == 72
 @test coefsCVmin == β[:,segCVmin]
 
-Random.seed!(13)
-coefsCV1se = coef(path, MinCV1se(path, 20))
-Random.seed!( 13)
-segCV1se = cross_validate_path(path,X,y, MinCV1se(path, 20),offset=offset)
-@test segCV1se == 42
+gen1 = Kfold(StableRNG(13), length(y), 20)
+coefsCV1se = coef(path, MinCV1se(gen1))
+gen2 = Kfold(StableRNG(13), length(y), 20)
+segCV1se = cross_validate_path(path, X, y, MinCV1se(gen2), offset=offset)
+@test segCV1se == 41
 @test coefsCV1se == β[:,segCV1se]
 
 end