Add missing and constant support for ZScore (#308)

* Add missing and constant support for ZScore

* Update test data

Author: juliohm

src/transforms/utils.jl

@@ -2,9 +2,9 @@
 # Licensed under the MIT License. See LICENSE in the project root.
 # ------------------------------------------------------------------
 
-zscore(x, μ, σ) = @. (x - μ) / σ
+zscore(x, μ, σ) = iszero(σ) ? fill(zero(μ), length(x)) : @. (x - μ) / σ
 
-revzscore(y, μ, σ) = @. σ * y + μ
+revzscore(y, μ, σ) = iszero(σ) ? fill(μ, length(y)) : @. σ * y + μ
 
 _assert(cond, msg) = cond || throw(AssertionError(msg))
 

src/transforms/zscore.jl

@@ -41,8 +41,8 @@ assertions(transform::ZScore) = [scitypeassert(Continuous, transform.selector)]
 isrevertible(::Type{<:ZScore}) = true
 
 function colcache(::ZScore, x)
-  μ = mean(x)
-  σ = std(x, mean=μ)
+  μ = mean(skipmissing(x))
+  σ = std(skipmissing(x), mean=μ)
   (μ=μ, σ=σ)
 end
 

test/data/eigenanalysis-2.png

@@ -87,4 +87,24 @@
   @test isapprox(std(n.b), 1; atol=1e-6)
   tₒ = revert(T, n, c)
   @test Tables.matrix(t) ≈ Tables.matrix(tₒ)
+
+  # missing values
+  a = [rand(Normal(7, 10), 99); missing]
+  t = Table(; a)
+  T = ZScore()
+  n, c = apply(T, t)
+  @test isapprox(mean(skipmissing(n.a)), 0; atol=1e-6)
+  @test isapprox(std(skipmissing(n.a)), 1; atol=1e-6)
+  tₒ = revert(T, n, c)
+  @test t.a[begin:(end - 1)] ≈ tₒ.a[begin:(end - 1)]
+  @test ismissing(tₒ.a[end])
+
+  # constant values
+  a = fill(2.0, 100)
+  t = Table(; a)
+  T = ZScore()
+  n, c = apply(T, t)
+  @test all(==(0), n.a)
+  tₒ = revert(T, n, c)
+  @test all(==(2), tₒ.a)
 end