From 36fe174c140b7cef7a58e0cf2fc3a5f2365be3d2 Mon Sep 17 00:00:00 2001
From: Dave Kleinschmidt <dave.f.kleinschmidt@gmail.com>
Date: Wed, 6 Dec 2017 16:00:41 -0500
Subject: [PATCH] modernize tests (#22)

* use testset for all tests
* use Compat.Test and import->using in tests
---
 test/REQUIRE                |   1 +
 test/conjugates.jl          | 214 +++++++++++++++++++-----------------
 test/conjugates_mvnormal.jl | 172 +++++++++++++++--------------
 test/conjugates_normal.jl   |   3 -
 test/runtests.jl            |  19 ++--
 5 files changed, 214 insertions(+), 195 deletions(-)
 create mode 100644 test/REQUIRE

diff --git a/test/REQUIRE b/test/REQUIRE
new file mode 100644
index 0000000..9530d2e
--- /dev/null
+++ b/test/REQUIRE
@@ -0,0 +1 @@
+Compat
\ No newline at end of file
diff --git a/test/conjugates.jl b/test/conjugates.jl
index f997dcf..cd70c64 100644
--- a/test/conjugates.jl
+++ b/test/conjugates.jl
@@ -1,4 +1,3 @@
-using Base.Test
 using Distributions
 using ConjugatePriors
 
@@ -26,144 +25,155 @@ function ccount(K, x, w)
 end
 
 
-# Beta - Bernoulli
+@testset "Beta - Bernoulli" begin
 
-pri = Beta(1.0, 2.0)
+    pri = Beta(1.0, 2.0)
 
-x = rand(Bernoulli(0.3), n)
-p = posterior(pri, Bernoulli, x)
-@test isa(p, Beta)
-@test p.α ≈ pri.α + sum(x)
-@test p.β ≈ pri.β + (n - sum(x))
+    x = rand(Bernoulli(0.3), n)
+    p = posterior(pri, Bernoulli, x)
+    @test isa(p, Beta)
+    @test p.α ≈ pri.α + sum(x)
+    @test p.β ≈ pri.β + (n - sum(x))
 
-f = fit_map(pri, Bernoulli, x)
-@test isa(f, Bernoulli)
-@test succprob(f) ≈ mode(p)
+    f = fit_map(pri, Bernoulli, x)
+    @test isa(f, Bernoulli)
+    @test succprob(f) ≈ mode(p)
 
-p = posterior(pri, Bernoulli, x, w)
-@test isa(p, Beta)
-@test p.α ≈ pri.α + sum(x .* w)
-@test p.β ≈ pri.β + (sum(w) - sum(x .* w))
+    p = posterior(pri, Bernoulli, x, w)
+    @test isa(p, Beta)
+    @test p.α ≈ pri.α + sum(x .* w)
+    @test p.β ≈ pri.β + (sum(w) - sum(x .* w))
 
-f = fit_map(pri, Bernoulli, x, w)
-@test isa(f, Bernoulli)
-@test succprob(f) ≈ mode(p)
+    f = fit_map(pri, Bernoulli, x, w)
+    @test isa(f, Bernoulli)
+    @test succprob(f) ≈ mode(p)
 
 
-# posterior_rand & posterior_randmodel
+    @testset "posterior_rand & posterior_randmodel" begin
 
-pri = Beta(1.0, 2.0)
-x = rand(Bernoulli(0.3), n)
-post = posterior(pri, Bernoulli, x)
+        pri = Beta(1.0, 2.0)
+        x = rand(Bernoulli(0.3), n)
+        post = posterior(pri, Bernoulli, x)
 
-pv = posterior_rand(pri, Bernoulli, x)
-@test isa(pv, Float64)
-@test 0. <= pv <= 1.
+        pv = posterior_rand(pri, Bernoulli, x)
+        @test isa(pv, Float64)
+        @test 0. <= pv <= 1.
 
-pv = posterior_rand(pri, Bernoulli, x, w)
-@test isa(pv, Float64)
-@test 0. <= pv <= 1.
+        pv = posterior_rand(pri, Bernoulli, x, w)
+        @test isa(pv, Float64)
+        @test 0. <= pv <= 1.
 
-pm = posterior_randmodel(pri, Bernoulli, x)
-@test isa(pm, Bernoulli)
-@test 0. <= succprob(pm) <= 1.
+        pm = posterior_randmodel(pri, Bernoulli, x)
+        @test isa(pm, Bernoulli)
+        @test 0. <= succprob(pm) <= 1.
 
-pm = posterior_randmodel(pri, Bernoulli, x, w)
-@test isa(pm, Bernoulli)
-@test 0. <= succprob(pm) <= 1.
+        pm = posterior_randmodel(pri, Bernoulli, x, w)
+        @test isa(pm, Bernoulli)
+        @test 0. <= succprob(pm) <= 1.
+    end
+    
+end
 
+@testset "Beta - Binomial" begin
 
-# Beta - Binomial
+    pri = Beta(1.0, 2.0)
 
-x = rand(Binomial(10, 0.3), n)
-p = posterior(pri, Binomial, (10, x))
-@test isa(p, Beta)
-@test p.α ≈ pri.α + sum(x)
-@test p.β ≈ pri.β + (10n - sum(x))
+    x = rand(Binomial(10, 0.3), n)
+    p = posterior(pri, Binomial, (10, x))
+    @test isa(p, Beta)
+    @test p.α ≈ pri.α + sum(x)
+    @test p.β ≈ pri.β + (10n - sum(x))
 
-f = fit_map(pri, Binomial, (10, x))
-@test isa(f, Binomial)
-@test ntrials(f) == 10
-@test succprob(f) ≈ mode(p)
+    f = fit_map(pri, Binomial, (10, x))
+    @test isa(f, Binomial)
+    @test ntrials(f) == 10
+    @test succprob(f) ≈ mode(p)
 
-p = posterior(pri, Binomial, (10, x), w)
-@test isa(p, Beta)
-@test p.α ≈ pri.α + sum(x .* w)
-@test p.β ≈ pri.β + (10 * sum(w) - sum(x .* w))
+    p = posterior(pri, Binomial, (10, x), w)
+    @test isa(p, Beta)
+    @test p.α ≈ pri.α + sum(x .* w)
+    @test p.β ≈ pri.β + (10 * sum(w) - sum(x .* w))
 
-f = fit_map(pri, Binomial, (10, x), w)
-@test isa(f, Binomial)
-@test ntrials(f) == 10
-@test succprob(f) ≈ mode(p)
+    f = fit_map(pri, Binomial, (10, x), w)
+    @test isa(f, Binomial)
+    @test ntrials(f) == 10
+    @test succprob(f) ≈ mode(p)
 
+end
 
-# Dirichlet - Categorical
+@testset "Dirichlet - Categorical" begin
 
-pri = Dirichlet([1., 2., 3.])
+    pri = Dirichlet([1., 2., 3.])
 
-x = rand(Categorical([0.2, 0.3, 0.5]), n)
-p = posterior(pri, Categorical, x)
-@test isa(p, Dirichlet)
-@test p.alpha ≈ pri.alpha + ccount(3, x)
+    x = rand(Categorical([0.2, 0.3, 0.5]), n)
+    p = posterior(pri, Categorical, x)
+    @test isa(p, Dirichlet)
+    @test p.alpha ≈ pri.alpha + ccount(3, x)
 
-f = fit_map(pri, Categorical, x)
-@test isa(f, Categorical)
-@test probs(f) ≈ mode(p)
+    f = fit_map(pri, Categorical, x)
+    @test isa(f, Categorical)
+    @test probs(f) ≈ mode(p)
 
-p = posterior(pri, Categorical, x, w)
-@test isa(p, Dirichlet)
-@test p.alpha ≈ pri.alpha + ccount(3, x, w)
+    p = posterior(pri, Categorical, x, w)
+    @test isa(p, Dirichlet)
+    @test p.alpha ≈ pri.alpha + ccount(3, x, w)
 
-f = fit_map(pri, Categorical, x, w)
-@test isa(f, Categorical)
-@test probs(f) ≈ mode(p)
+    f = fit_map(pri, Categorical, x, w)
+    @test isa(f, Categorical)
+    @test probs(f) ≈ mode(p)
 
+end
 
-# Dirichlet - Multinomial
+@testset "Dirichlet - Multinomial" begin
 
-x = rand(Multinomial(100, [0.2, 0.3, 0.5]), 1)
-p = posterior(pri, Multinomial, x)
-@test isa(p, Dirichlet)
-@test p.alpha ≈ pri.alpha + x
+    pri = Dirichlet([1., 2., 3.])
 
-r = posterior_mode(pri, Multinomial, x)
-@test r ≈ mode(p)
+    x = rand(Multinomial(100, [0.2, 0.3, 0.5]), 1)
+    p = posterior(pri, Multinomial, x)
+    @test isa(p, Dirichlet)
+    @test p.alpha ≈ pri.alpha + x
 
-x = rand(Multinomial(10, [0.2, 0.3, 0.5]), n)
-p = posterior(pri, Multinomial, x)
-@test isa(p, Dirichlet)
-@test p.alpha ≈ pri.alpha + vec(sum(x, 2))
+    r = posterior_mode(pri, Multinomial, x)
+    @test r ≈ mode(p)
 
-r = posterior_mode(pri, Multinomial, x)
-@test r ≈ mode(p)
+    x = rand(Multinomial(10, [0.2, 0.3, 0.5]), n)
+    p = posterior(pri, Multinomial, x)
+    @test isa(p, Dirichlet)
+    @test p.alpha ≈ pri.alpha + vec(sum(x, 2))
 
-p = posterior(pri, Multinomial, x, w)
-@test isa(p, Dirichlet)
-@test p.alpha ≈ pri.alpha + vec(x * w)
+    r = posterior_mode(pri, Multinomial, x)
+    @test r ≈ mode(p)
 
-r = posterior_mode(pri, Multinomial, x, w)
-@test r ≈ mode(p)
+    p = posterior(pri, Multinomial, x, w)
+    @test isa(p, Dirichlet)
+    @test p.alpha ≈ pri.alpha + vec(x * w)
 
+    r = posterior_mode(pri, Multinomial, x, w)
+    @test r ≈ mode(p)
 
-# Gamma - Exponential
+end
 
-pri = Gamma(1.5, 2.0)
+@testset "Gamma - Exponential" begin
 
-x = rand(Exponential(2.0), n)
-p = posterior(pri, Exponential, x)
-@test isa(p, Gamma)
-@test shape(p) ≈ shape(pri) + n
-@test rate(p) ≈ rate(pri) + sum(x)
+    pri = Gamma(1.5, 2.0)
 
-f = fit_map(pri, Exponential, x)
-@test isa(f, Exponential)
-@test rate(f) ≈ mode(p)
+    x = rand(Exponential(2.0), n)
+    p = posterior(pri, Exponential, x)
+    @test isa(p, Gamma)
+    @test shape(p) ≈ shape(pri) + n
+    @test rate(p) ≈ rate(pri) + sum(x)
 
-p = posterior(pri, Exponential, x, w)
-@test isa(p, Gamma)
-@test shape(p) ≈ shape(pri) + sum(w)
-@test rate(p) ≈ rate(pri) + sum(x .* w)
+    f = fit_map(pri, Exponential, x)
+    @test isa(f, Exponential)
+    @test rate(f) ≈ mode(p)
 
-f = fit_map(pri, Exponential, x, w)
-@test isa(f, Exponential)
-@test rate(f) ≈ mode(p)
+    p = posterior(pri, Exponential, x, w)
+    @test isa(p, Gamma)
+    @test shape(p) ≈ shape(pri) + sum(w)
+    @test rate(p) ≈ rate(pri) + sum(x .* w)
+
+    f = fit_map(pri, Exponential, x, w)
+    @test isa(f, Exponential)
+    @test rate(f) ≈ mode(p)
+
+end
diff --git a/test/conjugates_mvnormal.jl b/test/conjugates_mvnormal.jl
index 87bca41..51a08e3 100644
--- a/test/conjugates_mvnormal.jl
+++ b/test/conjugates_mvnormal.jl
@@ -2,115 +2,121 @@
 
 using Distributions
 using ConjugatePriors
-using Base.Test
 
-import ConjugatePriors: posterior, posterior_randmodel, NormalWishart, NormalInverseWishart
+using ConjugatePriors: posterior, posterior_randmodel, NormalWishart, NormalInverseWishart
 
-# MvNormal -- Normal (known covariance)
+@testset "Conjugate models for multivariate normal" begin
 
-n = 3
-p = 4
-X = reshape(Float64[1:12;], p, n)
-w = rand(n)
-Xw = X * diagm(w)
+    @testset "MvNormal -- Normal (known covariance)" begin
 
-Sigma = 0.75 * eye(p) + fill(0.25, 4, 4)
-ss = suffstats(MvNormalKnownCov(Sigma), X)
-ssw = suffstats(MvNormalKnownCov(Sigma), X, w)
+        n = 3
+        p = 4
+        X = reshape(Float64[1:12;], p, n)
+        w = rand(n)
+        Xw = X * diagm(w)
 
-s_t = sum(X, 2)
-ws_t = sum(Xw, 2)
-tw_t = length(w)
-wtw_t = sum(w)
+        Sigma = 0.75 * eye(p) + fill(0.25, 4, 4)
+        ss = suffstats(MvNormalKnownCov(Sigma), X)
+        ssw = suffstats(MvNormalKnownCov(Sigma), X, w)
 
-@test ss.sx ≈ s_t
-@test ss.tw ≈ tw_t
+        s_t = sum(X, 2)
+        ws_t = sum(Xw, 2)
+        tw_t = length(w)
+        wtw_t = sum(w)
 
-@test ssw.sx ≈ ws_t
-@test ssw.tw ≈ wtw_t
+        @test ss.sx ≈ s_t
+        @test ss.tw ≈ tw_t
 
-# Posterior
-n = 10
-# n = 100
-mu_true = [2., 3.]
-Sig_true = eye(2)
-Sig_true[1,2] = Sig_true[2,1] = 0.25
-mu0 = [2.5, 2.5]
-Sig0 = eye(2)
-Sig0[1,2] = Sig0[2,1] = 0.5
-X = rand(MultivariateNormal(mu_true, Sig_true), n)
-pri = MultivariateNormal(mu0, Sig0)
+        @test ssw.sx ≈ ws_t
+        @test ssw.tw ≈ wtw_t
 
-post = posterior((pri, Sig_true), MvNormal, X)
-@test isa(post, FullNormal)
+        # Posterior
+        n = 10
+        # n = 100
+        mu_true = [2., 3.]
+        Sig_true = eye(2)
+        Sig_true[1,2] = Sig_true[2,1] = 0.25
+        mu0 = [2.5, 2.5]
+        Sig0 = eye(2)
+        Sig0[1,2] = Sig0[2,1] = 0.5
+        X = rand(MultivariateNormal(mu_true, Sig_true), n)
+        pri = MultivariateNormal(mu0, Sig0)
 
-@test post.μ ≈ inv(inv(Sig0) + n*inv(Sig_true))*(n*inv(Sig_true)*mean(X,2) + inv(Sig0)*mu0)
-@test post.Σ.mat ≈ inv(inv(Sig0) + n*inv(Sig_true))
+        post = posterior((pri, Sig_true), MvNormal, X)
+        @test isa(post, FullNormal)
 
-# posterior_sample
+        @test post.μ ≈ inv(inv(Sig0) + n*inv(Sig_true))*(n*inv(Sig_true)*mean(X,2) + inv(Sig0)*mu0)
+        @test post.Σ.mat ≈ inv(inv(Sig0) + n*inv(Sig_true))
 
-ps = posterior_randmodel((pri, Sig_true), MvNormal, X)
-@test isa(ps, FullNormal)
-@test insupport(ps, ps.μ)
-@test insupport(InverseWishart, ps.Σ.mat)
+        # posterior_sample
 
+        ps = posterior_randmodel((pri, Sig_true), MvNormal, X)
+        @test isa(ps, FullNormal)
+        @test insupport(ps, ps.μ)
+        @test insupport(InverseWishart, ps.Σ.mat)
 
-# NormalInverseWishart - MvNormal
+    end
+    
+    @testset "NormalInverseWishart - MvNormal" begin
 
-mu_true = [2., 2.]
-Sig_true = eye(2)
-Sig_true[1,2] = Sig_true[2,1] = 0.25
+        mu_true = [2., 2.]
+        Sig_true = eye(2)
+        Sig_true[1,2] = Sig_true[2,1] = 0.25
 
-X = rand(MultivariateNormal(mu_true, Sig_true), n)
-Xbar = mean(X,2)
-Xm = X .- mean(X,2)
+        X = rand(MultivariateNormal(mu_true, Sig_true), n)
+        Xbar = mean(X,2)
+        Xm = X .- mean(X,2)
 
-mu0 = [2., 3.]
-kappa0 = 3.
-nu0 = 4.
-T0 = eye(2)
-T0[1,2] = T0[2,1] = .5
-pri = NormalInverseWishart(mu0, kappa0, T0, nu0)
+        mu0 = [2., 3.]
+        kappa0 = 3.
+        nu0 = 4.
+        T0 = eye(2)
+        T0[1,2] = T0[2,1] = .5
+        pri = NormalInverseWishart(mu0, kappa0, T0, nu0)
 
-post = posterior(pri, MvNormal, X)
+        post = posterior(pri, MvNormal, X)
 
-@test post.mu ≈ (kappa0*mu0 + n*Xbar)./(kappa0 + n)
-@test post.kappa ≈ kappa0 + n
-@test post.nu ≈ nu0 + n
-@test (post.Lamchol[:U]'*post.Lamchol[:U]) ≈ T0 + A_mul_Bt(Xm, Xm) + kappa0*n/(kappa0+n)*(Xbar-mu0)*(Xbar-mu0)'
+        @test post.mu ≈ (kappa0*mu0 + n*Xbar)./(kappa0 + n)
+        @test post.kappa ≈ kappa0 + n
+        @test post.nu ≈ nu0 + n
+        @test (post.Lamchol[:U]'*post.Lamchol[:U]) ≈ T0 + A_mul_Bt(Xm, Xm) + kappa0*n/(kappa0+n)*(Xbar-mu0)*(Xbar-mu0)'
 
-ps = posterior_randmodel(pri, MultivariateNormal, X)
+        ps = posterior_randmodel(pri, MultivariateNormal, X)
 
-@test isa(ps, MultivariateNormal)
-@test insupport(ps, ps.μ) && insupport(InverseWishart, ps.Σ.mat)
+        @test isa(ps, MultivariateNormal)
+        @test insupport(ps, ps.μ) && insupport(InverseWishart, ps.Σ.mat)
 
+    end
+    
+    @testset "NormalWishart - MvNormal" begin
 
-# NormalWishart - MvNormal
+        mu_true = [2., 2.]
+        Lam_true = eye(2)
+        Lam_true[1,2] = Lam_true[2,1] = 0.25
 
-mu_true = [2., 2.]
-Lam_true = eye(2)
-Lam_true[1,2] = Lam_true[2,1] = 0.25
+        X = rand(MvNormal(mu_true, inv(Lam_true)), n)
+        Xbar = mean(X,2)
+        Xm = X .- Xbar
 
-X = rand(MvNormal(mu_true, inv(Lam_true)), n)
-Xbar = mean(X,2)
-Xm = X .- Xbar
+        mu0 = [2., 3.]
+        kappa0 = 3.
+        nu0 = 4.
+        T0 = eye(2)
+        T0[1,2] = T0[2,1] = .5
+        pri = NormalWishart(mu0, kappa0, T0, nu0)
 
-mu0 = [2., 3.]
-kappa0 = 3.
-nu0 = 4.
-T0 = eye(2)
-T0[1,2] = T0[2,1] = .5
-pri = NormalWishart(mu0, kappa0, T0, nu0)
+        post = posterior(pri, MvNormal, X)
 
-post = posterior(pri, MvNormal, X)
+        @test post.mu ≈ (kappa0*mu0 + n*Xbar)./(kappa0 + n)
+        @test post.kappa ≈ kappa0 + n
+        @test post.nu ≈ nu0 + n
+        @test (post.Tchol[:U]'*post.Tchol[:U]) ≈ T0 + A_mul_Bt(Xm, Xm) + kappa0*n/(kappa0+n)*(Xbar-mu0)*(Xbar-mu0)'
+
+        ps = posterior_randmodel(pri, MvNormal, X)
 
-@test post.mu ≈ (kappa0*mu0 + n*Xbar)./(kappa0 + n)
-@test post.kappa ≈ kappa0 + n
-@test post.nu ≈ nu0 + n
-@test (post.Tchol[:U]'*post.Tchol[:U]) ≈ T0 + A_mul_Bt(Xm, Xm) + kappa0*n/(kappa0+n)*(Xbar-mu0)*(Xbar-mu0)'
+        @test isa(ps, MultivariateNormal)
+        @test insupport(ps, ps.μ)
+        @test insupport(InverseWishart, ps.Σ.mat)  # InverseWishart on purpose
 
-ps = posterior_randmodel(pri, MvNormal, X)
-
-@test isa(ps, MultivariateNormal)
-@test insupport(ps, ps.μ)
-@test insupport(InverseWishart, ps.Σ.mat)  # InverseWishart on purpose
+    end
+end
diff --git a/test/conjugates_normal.jl b/test/conjugates_normal.jl
index 6ed7a85..cd6bcb4 100644
--- a/test/conjugates_normal.jl
+++ b/test/conjugates_normal.jl
@@ -1,6 +1,3 @@
-# 
-
-using Base.Test
 using Distributions
 using ConjugatePriors
 
diff --git a/test/runtests.jl b/test/runtests.jl
index db98391..539272e 100644
--- a/test/runtests.jl
+++ b/test/runtests.jl
@@ -1,9 +1,14 @@
-tests = ["conjugates",
-         "conjugates_normal",
-         "conjugates_mvnormal"]
+using Compat
+using Compat.Test
 
-for t in tests
-    fpath = "$t.jl"
-    println("running $fpath ...")
-    include(fpath)
+@testset "Conjugate priors" begin
+    tests = ["conjugates",
+             "conjugates_normal",
+             "conjugates_mvnormal"]
+
+    for t in tests
+        fpath = "$t.jl"
+        println("running $fpath ...")
+        include(fpath)
+    end
 end