Clean MCMCChains, remove type piracy, and do not invalidate methods in Base

.travis.yml

@@ -1,11 +1,17 @@
 language: julia
 
+# avoid duplicate tests in PRs
+branches:
+  only:
+    - master
+
 os:
   - linux
 #  - osx
 
 julia:
   - 1.0
+  - 1.3
   - nightly
 
 matrix:
@@ -16,15 +22,8 @@ matrix:
 notifications:
   email: false
 
-#script:
-#  - if [[ -a .git/shallow ]]; then git fetch --unshallow; fi
-#  - julia -e "
-#        using Pkg;
-#        Pkg.clone(pwd(), \"MCMCChain\");
-#        Pkg.build(\"MCMCChain\");
-#        Pkg.test(\"MCMCChain\"; coverage=true)
-#      "
-
 after_success:
-  - julia -e 'using Pkg; cd(Pkg.dir("MCMCChain")); Pkg.add("Coverage"); using Coverage; Coveralls.submit(process_folder())'
-
+  - if [[ $TRAVIS_JULIA_VERSION = 1.3 ]] && [[ $TRAVIS_OS_NAME = linux ]]; then
+      julia -e 'using Pkg; Pkg.add("Coverage"); using Coverage; Codecov.submit(process_folder())';
+      julia -e 'using Pkg; Pkg.add("Coverage"); using Coverage; Coveralls.submit(process_folder())';
+    fi

Project.toml

@@ -10,12 +10,10 @@ AbstractMCMC = "80f14c24-f653-4e6a-9b94-39d6b0f70001"
 AxisArrays = "39de3d68-74b9-583c-8d2d-e117c070f3a9"
 DataFrames = "a93c6f00-e57d-5684-b7b6-d8193f3e46c0"
 Distributions = "31c24e10-a181-5473-b8eb-7969acd0382f"
-KernelDensity = "5ab0869b-81aa-558d-bb23-cbf5423bbe9b"
 LinearAlgebra = "37e2e46d-f89d-539d-b4ee-838fcccc9c8e"
 Random = "9a3f8284-a2c9-5f02-9a11-845980a1fd5c"
 RecipesBase = "3cdcf5f2-1ef4-517c-9805-6587b60abb01"
 Serialization = "9e88b42a-f829-5b0c-bbe9-9e923198166b"
-Showoff = "992d4aef-0814-514b-bc4d-f2e9a6c4116f"
 SpecialFunctions = "276daf66-3868-5448-9aa4-cd146d93841b"
 Statistics = "10745b16-79ce-11e8-11f9-7d13ad32a3b2"
 StatsBase = "2913bbd2-ae8a-5f71-8c99-4fb6c76f3a91"
@@ -25,18 +23,16 @@ AbstractMCMC = "~0.1, 0.2, 0.3"
 AxisArrays = "^0.3, 0.4"
 DataFrames = "^0.19, 0.20"
 Distributions = "^0.21, 0.22"
-KernelDensity = "^0.5"
 RecipesBase = "^0.7, 0.8"
-Showoff = "^0.3"
 SpecialFunctions = "^0.8, 0.9, 0.10"
 StatsBase = "^0.32"
-Turing = "^0.7"
 julia = "^1"
 
 [extras]
+KernelDensity = "5ab0869b-81aa-558d-bb23-cbf5423bbe9b"
 StatsPlots = "f3b207a7-027a-5e70-b257-86293d7955fd"
 Test = "8dfed614-e22c-5e08-85e1-65c5234f0b40"
 Turing = "fce5fe82-541a-59a6-adf8-730c64b5f9a0"
 
 [targets]
-test = ["StatsPlots", "Test", "Turing"]
+test = ["KernelDensity", "StatsPlots", "Test", "Turing"]

README.md

@@ -11,24 +11,23 @@ The following simple example illustrates how to use Chain to visually summarize
 using MCMCChains
 using StatsPlots
 
-theme(:ggplot2);
+theme(:ggplot2)
 
 # Define the experiment
-n_iter = 500;
-n_name = 3;
-n_chain = 2;
+n_iter = 500
+n_name = 3
+n_chain = 2
 
 # experiment results
-val = randn(n_iter, n_name, n_chain) .+ [1, 2, 3]';
-val = hcat(val, rand(1:2, n_iter, 1, n_chain));
+val = randn(n_iter, n_name, n_chain) .+ [1, 2, 3]'
+val = hcat(val, rand(1:2, n_iter, 1, n_chain))
 
 # construct a Chains object
-chn = Chains(val);
+chn = Chains(val)
 
 # visualize the MCMC simulation results
 p1 = plot(chn)
 p2 = plot(chn, colordim = :parameter)
-
 ```
 This code results in the visualizations shown below. Note that the plot function takes the additional arguments described in the [Plots.jl](https://github.com/JuliaPlots/Plots.jl) package.
 
@@ -176,27 +175,27 @@ Note that `x.P` is a tuple which has to be indexed by the relevant index, while
 Options for method are  `[:weiss, :hangartner, :DARBOOT, MCBOOT, :billinsgley, :billingsleyBOOT]`
 
 ```julia
-discretediag(c::AbstractChains; frac=0.3, method=:weiss, nsim=1000)
+discretediag(c::Chains; frac=0.3, method=:weiss, nsim=1000)
 ```
 
 #### Gelman, Rubin, and Brooks Diagnostics
 ```julia
-gelmandiag(c::AbstractChains; alpha=0.05, mpsrf=false, transform=false)
+gelmandiag(c::Chains; alpha=0.05, mpsrf=false, transform=false)
 ```
 
 #### Geweke Diagnostic
 ```julia
-gewekediag(c::AbstractChains; first=0.1, last=0.5, etype=:imse)
+gewekediag(c::Chains; first=0.1, last=0.5, etype=:imse)
 ```
 
 #### Heidelberger and Welch Diagnostics
 ```julia
-heideldiag(c::AbstractChains; alpha=0.05, eps=0.1, etype=:imse)
+heideldiag(c::Chains; alpha=0.05, eps=0.1, etype=:imse)
 ```
 
 #### Raftery and Lewis Diagnostic
 ```julia
-rafterydiag(c::AbstractChains; q=0.025, r=0.005, s=0.95, eps=0.001)
+rafterydiag(c::Chains; q=0.025, r=0.005, s=0.95, eps=0.001)
 ```
 
 ### Model Selection
@@ -217,30 +216,31 @@ DIC, pD = dic(chn, lpfun)
 ### Plotting
 ```julia
 # construct a plot
-plot(c::AbstractChains, seriestype = (:traceplot, :mixeddensity))
+plot(c::Chains, seriestype = (:traceplot, :mixeddensity))
 
 # construct trace plots
-plot(c::AbstractChains, seriestype = :traceplot)
+plot(c::Chains, seriestype = :traceplot)
+
 # or for all seriestypes use the alternative shorthand syntax
-traceplot(c::AbstractChains)
+traceplot(c::Chains)
 
 # construct running average plots
-meanplot(c::AbstractChains)
+meanplot(c::Chains)
 
 # construct density plots
-density(c::AbstractChains)
+density(c::Chains)
 
 # construct histogram plots
-histogram(c::AbstractChains)
+histogram(c::Chains)
 
 # construct mixed density plots
-mixeddensity(c::AbstractChains)
+mixeddensity(c::Chains)
 
 # construct autocorrelation plots
-autocorplot(c::AbstractChains)
+autocorplot(c::Chains)
 
 # make a cornerplot (requires StatPlots) of parameters in a Chain:
-corner(c::AbstractChains, [:A, :B])
+corner(c::Chains, [:A, :B])
 ```
 
 ### Saving and Loading Chains
@@ -277,7 +277,7 @@ Array(chns, append_chains=false)
 Array(chns, remove_missing_union=false)
 
 # This will not convert the Array columns from a
-`Union{Missing, Real}` to a `Vector{Real}`.
+# `Union{Missing, Real}` to a `Vector{Real}`.
 ```
 
 Similarly, for DataFrames:
@@ -291,27 +291,34 @@ DataFrame(chns, append_chains=false)
 DataFrame(chns, remove_missing_union=false)
 ```
 
-See also ?MCMCChains.DataFrame and ?MCMCChains.Array for more help.
+See also `?DataFrame` and `?Array` for more help.
 
 ### Sampling Chains
 
-MCMCChains overloads several `sample()` methods as defined in StatsBase:
+MCMCChains overloads several `sample` methods as defined in StatsBase:
 
 ```julia
 # Sampling `n` samples from the chain `a`. Optionally
 # weighting the samples using `wv`.
 sample([rng], a, [wv::AbstractWeights], n::Integer)
 
-# E.g. creating 10000 weighted samples:
-c = kde(Array(chn[:s]))
-chn_weighted_sample = sample(c.x, Weights(c.density), 100000)
-
 # As above, but supports replacing and ordering.
 sample([rng], a, [wv::AbstractWeights], n::Integer; replace=true,
   ordered=false)
 ```
 
-See also ?MCMCChains.sample for additional help.
+See also `?sample` for additional help. Alternatively, you can construct
+and sample from a kernel density estimator using the KernelDensity package:
+
+```julia
+using KernelDensity
+
+# Construct a kernel density estimator
+c = kde(Array(chn[:s]))
+
+# Generate 10000 weighted samples from the grid points
+chn_weighted_sample = sample(c.x, Weights(c.density), 100000)
+```
 
 ## License Notice
 Note that this package heavily uses and adapts code from the Mamba.jl package licensed under MIT License, see License.md.

src/MCMCChains.jl

@@ -1,39 +1,32 @@
 module MCMCChains
 
-using AbstractMCMC
-
-import Statistics
-import Showoff: showoff
-import StatsBase: autocor, autocov, countmap, counts, describe, predict,
-       quantile, sample, sem, summarystats, sample, AbstractWeights
-import LinearAlgebra: diag
-import Serialization: serialize, deserialize
-import Base: sort, range, names, get, hash, convert, show, display
-import Statistics: cor, mean
-import Core.Array
-import DataFrames: DataFrame, names, eachcol
-
-using RecipesBase
-import RecipesBase: plot
+using AxisArrays
+const axes = Base.axes
 
-using Serialization
+using AbstractMCMC
+using DataFrames: eachcol
+import DataFrames: DataFrame
 using Distributions
-using KernelDensity
+using RecipesBase
 using SpecialFunctions
-using AxisArrays
-const axes = Base.axes
+using StatsBase: autocov, counts, sem, AbstractWeights
+import StatsBase: autocor, describe, quantile, sample, summarystats
 
-export Chains, getindex, setindex!, chains, setinfo, chainscat
-export describe, set_section, get_params, sections, set_names
-export sample, AbstractWeights
-export Array, DataFrame, sort_sections, convert
-export summarize, summarystats, ChainDataFrame
-export hpd, ess
+using LinearAlgebra: diag
+import Serialization: serialize, deserialize
+import Random
+import Statistics: std, cor, mean
+
+export Chains, chains, chainscat
+export set_section, get_params, sections, sort_sections, setinfo, set_names
 export mean
+export autocor, describe, sample, summarystats, AbstractWeights
+export ChainDataFrame, DataFrame
+export summarize
 
 # export diagnostics functions
 export discretediag, gelmandiag, gewekediag, heideldiag, rafterydiag
-export autocor
+export hpd, ess
 
 """
     Chains type
@@ -53,9 +46,7 @@ struct Chains{A, T, K<:NamedTuple, L<:NamedTuple} <: AbstractChains
     info::L
 end
 
-# imports
 include("utils.jl")
-
 include("chains.jl")
 include("constructors.jl")
 include("summarize.jl")