Skip to content

Preserve eltype where possible for moments #688

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
wants to merge 9 commits into
base: master
Choose a base branch
from
Open

Preserve eltype where possible for moments #688

Show file tree
Hide file tree
Changes from 5 commits
Commits
Show all changes
9 commits
Select commit Hold shift + click to select a range
b54b9fa
Preserve eltype where possible for moments
palday Apr 23, 2021
f85a002
rm extra newline
palday Apr 23, 2021
94d2656
simplify
palday Apr 26, 2021
56e0fff
get serious about promotion
palday Apr 26, 2021
40e7290
flatten
palday Apr 26, 2021
338eb05
even more type stable
palday Apr 26, 2021
f05e14f
Apply suggestions from code review
palday Apr 28, 2021
eb0e7c3
similar type stability
palday Apr 28, 2021
aacef2a
skewness and kurtosis
palday May 4, 2021
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
56 changes: 35 additions & 21 deletions src/moments.jl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -42,7 +42,7 @@ function var(v::RealArray, w::AbstractWeights; mean=nothing,
corrected::DepBool=nothing)
corrected = depcheck(:var, corrected)

if mean == nothing
if mean === nothing
varm(v, w, Statistics.mean(v, w); corrected=corrected)
else
varm(v, w, mean; corrected=corrected)
Expand Down Expand Up @@ -85,14 +85,16 @@ end
function varm(A::RealArray, w::AbstractWeights, M::RealArray, dim::Int;
corrected::DepBool=nothing)
corrected = depcheck(:varm, corrected)
varm!(similar(A, Float64, Base.reduced_indices(axes(A), dim)), A, w, M,
T = promote_type(eltype(A), eltype(w), eltype(M))
varm!(similar(A, T, Base.reduced_indices(axes(A), dim)), A, w, M,
dim; corrected=corrected)
end

function var(A::RealArray, w::AbstractWeights, dim::Int; mean=nothing,
corrected::DepBool=nothing)
corrected = depcheck(:var, corrected)
var!(similar(A, Float64, Base.reduced_indices(axes(A), dim)), A, w, dim;
T = promote_type(eltype(A), eltype(w))
var!(similar(A, T, Base.reduced_indices(axes(A), dim)), A, w, dim;
mean=mean, corrected=corrected)
end

Expand Down Expand Up @@ -223,7 +225,8 @@ end
##### General central moment
function _moment2(v::RealArray, m::Real; corrected=false)
n = length(v)
s = 0.0
T = promote_type(eltype(v), typeof(m))
s = zero(T)
for i = 1:n
@inbounds z = v[i] - m
s += z * z
Expand All @@ -233,7 +236,8 @@ end

function _moment2(v::RealArray, wv::AbstractWeights, m::Real; corrected=false)
n = length(v)
s = 0.0
T = promote_type(eltype(v), eltype(wv), typeof(m))
s = zero(T)
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

To ensure type stability, how about following exactly what the loop does and using T = typeof(zero(eltype(v) - zero(m))^2 * zero(eltype(w)))? That's what I did in my first attempt to moving these functions to Statistics: https://github.com/JuliaLang/julia/pull/31395/files#diff-6b82a0c4e60cbe68048e35375987a86688afb4aea60dda8f2d2d4938dc845776R375

for i = 1:n
@inbounds z = v[i] - m
@inbounds s += (z * z) * wv[i]
Expand All @@ -244,7 +248,8 @@ end

function _moment3(v::RealArray, m::Real)
n = length(v)
s = 0.0
T = promote_type(eltype(v), typeof(m))
s = zero(T)
for i = 1:n
@inbounds z = v[i] - m
s += z * z * z
Expand All @@ -254,7 +259,8 @@ end

function _moment3(v::RealArray, wv::AbstractWeights, m::Real)
n = length(v)
s = 0.0
T = promote_type(eltype(v), eltype(wv), typeof(m))
s = zero(T)
for i = 1:n
@inbounds z = v[i] - m
@inbounds s += (z * z * z) * wv[i]
Expand All @@ -264,7 +270,8 @@ end

function _moment4(v::RealArray, m::Real)
n = length(v)
s = 0.0
T = promote_type(eltype(v), typeof(m))
s = zero(T)
for i = 1:n
@inbounds z = v[i] - m
s += abs2(z * z)
Expand All @@ -274,7 +281,8 @@ end

function _moment4(v::RealArray, wv::AbstractWeights, m::Real)
n = length(v)
s = 0.0
T = promote_type(eltype(v), eltype(wv), typeof(m))
s = zero(T)
for i = 1:n
@inbounds z = v[i] - m
@inbounds s += abs2(z * z) * wv[i]
Expand All @@ -284,7 +292,8 @@ end

function _momentk(v::RealArray, k::Int, m::Real)
n = length(v)
s = 0.0
T = promote_type(eltype(v), typeof(m))
s = zero(T)
for i = 1:n
@inbounds z = v[i] - m
s += (z ^ k)
Expand All @@ -294,7 +303,8 @@ end

function _momentk(v::RealArray, k::Int, wv::AbstractWeights, m::Real)
n = length(v)
s = 0.0
T = promote_type(eltype(v), eltype(wv), typeof(m))
s = zero(T)
for i = 1:n
@inbounds z = v[i] - m
@inbounds s += (z ^ k) * wv[i]
Expand Down Expand Up @@ -341,8 +351,9 @@ specifying a weighting vector `wv` and a center `m`.
"""
function skewness(v::RealArray, m::Real)
n = length(v)
cm2 = 0.0 # empirical 2nd centered moment (variance)
cm3 = 0.0 # empirical 3rd centered moment
T = promote_type(eltype(v), typeof(m))
cm2 = zero(T) # empirical 2nd centered moment (variance)
cm3 = zero(T) # empirical 3rd centered moment
for i = 1:n
@inbounds z = v[i] - m
z2 = z * z
Expand All @@ -358,8 +369,9 @@ end
function skewness(v::RealArray, wv::AbstractWeights, m::Real)
n = length(v)
length(wv) == n || throw(DimensionMismatch("Inconsistent array lengths."))
cm2 = 0.0 # empirical 2nd centered moment (variance)
cm3 = 0.0 # empirical 3rd centered moment
T = promote_type(eltype(v), eltype(wv), typeof(m))
cm2 = zero(T) # empirical 2nd centered moment (variance)
cm3 = zero(T) # empirical 3rd centered moment

@inbounds for i = 1:n
x_i = v[i]
Expand Down Expand Up @@ -388,8 +400,9 @@ specifying a weighting vector `wv` and a center `m`.
"""
function kurtosis(v::RealArray, m::Real)
n = length(v)
cm2 = 0.0 # empirical 2nd centered moment (variance)
cm4 = 0.0 # empirical 4th centered moment
T = promote_type(eltype(v), typeof(m))
cm2 = zero(T) # empirical 2nd centered moment (variance)
cm4 = zero(T) # empirical 4th centered moment
for i = 1:n
@inbounds z = v[i] - m
z2 = z * z
Expand All @@ -398,14 +411,15 @@ function kurtosis(v::RealArray, m::Real)
end
cm4 /= n
cm2 /= n
return (cm4 / (cm2 * cm2)) - 3.0
return (cm4 / (cm2 * cm2)) - 3
end

function kurtosis(v::RealArray, wv::AbstractWeights, m::Real)
n = length(v)
length(wv) == n || throw(DimensionMismatch("Inconsistent array lengths."))
cm2 = 0.0 # empirical 2nd centered moment (variance)
cm4 = 0.0 # empirical 4th centered moment
T = promote_type(eltype(v), eltype(wv), typeof(m))
cm2 = zero(T) # empirical 2nd centered moment (variance)
cm4 = zero(T) # empirical 4th centered moment

@inbounds for i = 1 : n
x_i = v[i]
Expand All @@ -419,7 +433,7 @@ function kurtosis(v::RealArray, wv::AbstractWeights, m::Real)
sw = sum(wv)
cm4 /= sw
cm2 /= sw
return (cm4 / (cm2 * cm2)) - 3.0
return (cm4 / (cm2 * cm2)) - 3
end

kurtosis(v::RealArray) = kurtosis(v, mean(v))
Expand Down
16 changes: 16 additions & 0 deletions test/moments.jl
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,4 +1,5 @@
using StatsBase
using Statistics
using Test

@testset "StatsBase.Moments" begin
Expand Down Expand Up @@ -278,4 +279,19 @@ end
@test moment(x, 5, w) ≈ sum((x2 .- 4).^5) / 5
end

@testset "Preservation of eltypes in moments" begin
xs = Float16[1, 2, 3, 4, 5];
ws = AnalyticWeights(Float16[1, 1, 1, 1, 1]);
@test typeof(std(xs)) === Float16
@test typeof(var(xs)) === Float16
@test typeof(mean(xs, ws)) === Float16
@test typeof(std(xs, ws)) === Float16
@test typeof(var(xs, ws)) === Float16
@test typeof(skewness(xs, ws)) === Float16
@test typeof(kurtosis(xs, ws)) === Float16
for i in 1:5
@test typeof(moment(xs, i, ws)) === Float16
end
end

end # @testset "StatsBase.Moments"