Support MCMCDiagnosticTools v0.2 (#392)

* Create permutedims utility function

* Bump compat

* Permute dims

* Don't worry about building y

* Remove unnecessary version checks

* Remove Compat dependency for eachcols

* Bump minimum tested version

* Bump minor version

* Test against permuted dims

* Add undocumented sections keyword arg

* Bump test min Julia version to v1.3

* Set Julia lower bound to 1.6

* Update minimum tested version

.github/workflows/CI.yml

@@ -13,7 +13,7 @@ jobs:
     strategy:
       matrix:
         version:
-          - '1.0'
+          - '1.6'
           - '1'
           - nightly
         os:

Project.toml

@@ -3,12 +3,11 @@ uuid = "c7f686f2-ff18-58e9-bc7b-31028e88f75d"
 keywords = ["markov chain monte carlo", "probablistic programming"]
 license = "MIT"
 desc = "Chain types and utility functions for MCMC simulations."
-version = "5.5.0"
+version = "5.6.0"
 
 [deps]
 AbstractMCMC = "80f14c24-f653-4e6a-9b94-39d6b0f70001"
 AxisArrays = "39de3d68-74b9-583c-8d2d-e117c070f3a9"
-Compat = "34da2185-b29b-5c13-b0c7-acf172513d20"
 Dates = "ade2ca70-3891-5945-98fb-dc099432e06a"
 Distributions = "31c24e10-a181-5473-b8eb-7969acd0382f"
 Formatting = "59287772-0a20-5a39-b81b-1366585eb4c0"
@@ -32,12 +31,11 @@ Tables = "bd369af6-aec1-5ad0-b16a-f7cc5008161c"
 [compat]
 AbstractMCMC = "0.4, 0.5, 1.0, 2.0, 3.0, 4"
 AxisArrays = "0.4.4"
-Compat = "2.2, 3, 4"
 Distributions = "0.21, 0.22, 0.23, 0.24, 0.25"
 Formatting = "0.4"
 IteratorInterfaceExtensions = "0.1.1, 1"
 KernelDensity = "0.6.2"
-MCMCDiagnosticTools = "0.1"
+MCMCDiagnosticTools = "0.2"
 MLJModelInterface = "0.3.5, 0.4, 1.0"
 NaturalSort = "1"
 OrderedCollections = "1.4"
@@ -47,4 +45,4 @@ StatsBase = "0.32, 0.33"
 StatsFuns = "0.8, 0.9, 1"
 TableTraits = "0.4, 1"
 Tables = "1"
-julia = "1"
+julia = "1.6"

src/MCMCChains.jl

@@ -4,7 +4,6 @@ using AxisArrays
 const axes = Base.axes
 import AbstractMCMC
 import AbstractMCMC: chainscat
-using Compat
 using Distributions
 using RecipesBase
 using Formatting
@@ -91,26 +90,13 @@ include("rstar.jl")
 # so we use the following hack
 const _read = Base.read
 const _write = Base.write
-@static if VERSION < v"1.1"
-    Base.@deprecate _read(
-        f::AbstractString,
-        ::Type{T}
-    ) where {T<:Chains} open(Serialization.deserialize, f, "r") false
-    Base.@deprecate _write(
-        f::AbstractString,
-        c::Chains
-    ) open(f, "w") do io
-        Serialization.serialize(io, c)
-    end false
-else
-    Base.@deprecate _read(
-        f::AbstractString,
-        ::Type{T}
-    ) where {T<:Chains} Serialization.deserialize(f) false
-    Base.@deprecate _write(
-        f::AbstractString,
-        c::Chains
-    ) Serialization.serialize(f, c) false
-end
+Base.@deprecate _read(
+    f::AbstractString,
+    ::Type{T}
+) where {T<:Chains} Serialization.deserialize(f) false
+Base.@deprecate _write(
+    f::AbstractString,
+    c::Chains
+) Serialization.serialize(f, c) false
 
 end # module

src/discretediag.jl

@@ -14,7 +14,7 @@ function MCMCDiagnosticTools.discretediag(
 
     # Compute statistics.
     between_chain_vals, within_chain_vals = MCMCDiagnosticTools.discretediag(
-        _chains.value.data; kwargs...
+        _permutedims_diagnostics(_chains.value.data); kwargs...
     )
 
     # Create dataframes

src/ess.jl

@@ -14,7 +14,10 @@ function MCMCDiagnosticTools.ess_rhat(
     _chains = Chains(chains, _clean_sections(chains, sections))
 
     # Estimate the effective sample size and rhat
-    ess, rhat = MCMCDiagnosticTools.ess_rhat(_chains.value.data; kwargs...)
+    ess, rhat = MCMCDiagnosticTools.ess_rhat(
+        _permutedims_diagnostics(_chains.value.data);
+        kwargs...,
+    )
 
     # Calculate ESS/minute if available
     dur = duration(chains)

src/gelmandiag.jl

@@ -9,7 +9,7 @@ function MCMCDiagnosticTools.gelmandiag(
 
     # Compute the potential scale reduction factor.
     psi = transform ? link(_chains) : _chains.value.data
-    results = MCMCDiagnosticTools.gelmandiag(psi; kwargs...)
+    results = MCMCDiagnosticTools.gelmandiag(_permutedims_diagnostics(psi); kwargs...)
 
     # Create a data frame with the results.
     df = ChainDataFrame(
@@ -31,7 +31,10 @@ function MCMCDiagnosticTools.gelmandiag_multivariate(
 
     # Compute the potential scale reduction factor.
     psi = transform ? link(_chains) : _chains.value.data
-    results = MCMCDiagnosticTools.gelmandiag_multivariate(psi; kwargs...)
+    results = MCMCDiagnosticTools.gelmandiag_multivariate(
+        _permutedims_diagnostics(psi);
+        kwargs...,
+    )
 
     # Create a data frame with the results.
     df = ChainDataFrame(

src/rstar.jl

@@ -44,14 +44,17 @@ function MCMCDiagnosticTools.rstar(
 end
 
 function MCMCDiagnosticTools.rstar(
-    rng::Random.AbstractRNG, classif::MLJModelInterface.Supervised, chn::Chains; kwargs...
+    rng::Random.AbstractRNG,
+    classif::MLJModelInterface.Supervised,
+    chn::Chains;
+    sections = _default_sections(chn),
+    kwargs...
 )
     nchains = size(chn, 3)
     nchains <= 1 && throw(DimensionMismatch())
 
-    # collect data
-    x = Array(chn)
-    y = repeat(chains(chn); inner = size(chn,1))
+    _chn = Chains(chn, _clean_sections(chn, sections))
+    x = _permutedims_diagnostics(_chn.value.data)
 
-    return MCMCDiagnosticTools.rstar(rng, classif, x, y; kwargs...)
+    return MCMCDiagnosticTools.rstar(rng, classif, x; kwargs...)
 end

src/utils.jl

@@ -197,3 +197,6 @@ function _isstrictlyincreasing_nonempty(x::AbstractVector{Int})
     end
     return true
 end
+
+# permute dims to match the dimension order of MCMCDiagnosticsTools
+_permutedims_diagnostics(x) = PermutedDimsArray(x, (1, 3, 2))

test/Project.toml

@@ -38,4 +38,4 @@ StatsPlots = "0.14.17, 0.15"
 TableTraits = "1"
 Tables = "1.3.1"
 UnicodePlots = "2, 3"
-julia = "1"
+julia = "1.6"

test/ess_tests.jl

@@ -30,7 +30,7 @@ end
         ess_df = ess_rhat(chain; method = method)
 
         # analyze array
-        ess_array, rhat_array = ess_rhat(x; method = method)
+        ess_array, rhat_array = ess_rhat(permutedims(x, (1, 3, 2)); method = method)
 
         @test ess_df[:,2] == ess_array
         @test ess_df[:,3] == rhat_array
@@ -46,7 +46,7 @@ end
         ess_df = ess_rhat(chain; method = method)
 
         # analyze array
-        ess_array, rhat_array = ess_rhat(val; method = method)
+        ess_array, rhat_array = ess_rhat(permutedims(val, (1, 3, 2)); method = method)
 
         @test ismissing(ess_df[:,2][1]) # since min(maxlag, niter - 1) = 0
         @test ismissing(ess_df[:,3][1])

test/runtests.jl

@@ -1,11 +1,3 @@
-# Activate test environment on older Julia versions
-if VERSION < v"1.2"
-    using Pkg
-    Pkg.activate(@__DIR__)
-    Pkg.develop(PackageSpec(path=dirname(@__DIR__)))
-    Pkg.instantiate()
-end
-
 using MCMCChains
 using Documenter