ci: test stan models (#54)

* ci: test stan models * feat(stan): kidiq.data.json * feat(stan): wells.data.json * feat(stan): esoph.data.json * feat(stan): roaches.data.json * feat(stan): cheese.data.json * ci: fix stan data command * fix(stan): 02-linear_regression_QR * fix(stan): 08-robust_robit_regression * fix(stan): 10-robust_zero_inflated_regression poisson opt * fix(stan): invalid operator in 11 ncp * fix(stan): 01 prior/posterior checks data file * fix(readme): note on CI
storopoli · Dec 3, 2023 · cf4aa81 · cf4aa81
1 parent 92b6fbc
commit cf4aa81
Show file tree

Hide file tree

Showing 14 changed files with 7,329 additions and 19 deletions.
diff --git a/.github/workflows/models.yml b/.github/workflows/models.yml
@@ -45,3 +45,111 @@ jobs:
         run: |
           julia --project=turing -e 'using Pkg; Pkg.instantiate()'
           julia --project=turing turing/${{ matrix.jl-file }}
+
+  test-stan:
+    name: Test Stan Models
+    runs-on: ubuntu-latest
+    strategy:
+      matrix:
+        stan:
+          [
+            {
+              model: "01-predictive_checks-posterior",
+              data: "coin_flip.data.json",
+            },
+            {
+              model: "01-predictive_checks-prior",
+              data: "coin_flip.data.json",
+            },
+            { model: "02-linear_regression-QR", data: "kidiq.data.json" },
+            {
+              model: "02-linear_regression-optimized",
+              data: "kidiq.data.json",
+            },
+            { model: "02-linear_regression", data: "kidiq.data.json" },
+            {
+              model: "03-logistic_regression-optimized",
+              data: "wells.data.json",
+            },
+            { model: "03-logistic_regression", data: "wells.data.json" },
+            {
+              model: "04-ordinal_regression-optimized",
+              data: "esoph.data.json",
+            },
+            { model: "04-ordinal_regression", data: "esoph.data.json" },
+            {
+              model: "05-poisson_regression-optimized",
+              data: "roaches.data.json",
+            },
+            { model: "05-poisson_regression", data: "roaches.data.json" },
+            { model: "06-robust_linear_regression", data: "kidiq.data.json" },
+            {
+              model: "07-robust_beta_binomial_regression",
+              data: "wells.data.json",
+            },
+            { model: "08-robust_robit_regression", data: "wells.data.json" },
+            {
+              model: "09-robust_negative_binomial_regression-optimized",
+              data: "roaches.data.json",
+            },
+            {
+              model: "09-robust_negative_binomial_regression",
+              data: "roaches.data.json",
+            },
+            {
+              model: "10-robust_zero_inflated_regression-negative_binomial-optimized",
+              data: "roaches.data.json",
+            },
+            {
+              model: "10-robust_zero_inflated_regression-negative_binomial",
+              data: "roaches.data.json",
+            },
+            {
+              model: "10-robust_zero_inflated_regression-poisson-optimized",
+              data: "roaches.data.json",
+            },
+            {
+              model: "10-robust_zero_inflated_regression-poisson",
+              data: "roaches.data.json",
+            },
+            {
+              model: "11-hierarchical_varying_intercept-non_centered",
+              data: "cheese.data.json",
+            },
+            {
+              model: "11-hierarchical_varying_intercept",
+              data: "cheese.data.json",
+            },
+            {
+              model: "12-hierarchical_varying_intercept_slope",
+              data: "cheese.data.json",
+            },
+            {
+              model: "13-model_comparison-negative_binomial",
+              data: "roaches.data.json",
+            },
+            { model: "13-model_comparison-poisson", data: "roaches.data.json" },
+            {
+              model: "13-model_comparison-zero_inflated-negative_binomial",
+              data: "roaches.data.json",
+            },
+            {
+              model: "13-model_comparison-zero_inflated-poisson",
+              data: "roaches.data.json",
+            },
+          ]
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@v4
+
+      - name: Install Nix
+        uses: DeterminateSystems/nix-installer-action@v8
+
+      - name: Cache Nix artifacts
+        uses: DeterminateSystems/magic-nix-cache-action@v2
+
+      - name: Test ${{ matrix.stan.model }}
+        run: |
+          echo "Compiling: ${{ matrix.stan.model }}"
+          nix develop -L . --command bash -c "stan stan/${{ matrix.stan.model }}"
+          nix develop -L . --command bash -c "stan/${{ matrix.stan.model }} sample data file=stan/${{ matrix.stan.data }}"
diff --git a/README.md b/README.md
@@ -29,23 +29,26 @@ The whole content is a set of several slides found at [`the latest release`](htt
 Here is a brief table of contents:
 
 1. **What is Bayesian Statistics?**
-2. **Common Probability Distributions**
-3. **Priors**
-4. **Predictive Checks**
-5. **Bayesian Linear Regression**
-6. **Bayesian Logistic Regression**
-7. **Bayesian Ordinal Regression**
-8. **Bayesian Regression with Count Data: Poisson Regression**
-9. **Robust Bayesian Regression**
-10. **Hierarchical Models**
-11. **Markov Chain Monte Carlo (MCMC) and Model Metrics**
-12. **Model Comparison: Cross-Validation and Other Metrics**
+1. **Common Probability Distributions**
+1. **Priors**
+1. **Predictive Checks**
+1. **Bayesian Linear Regression**
+1. **Bayesian Logistic Regression**
+1. **Bayesian Ordinal Regression**
+1. **Bayesian Regression with Count Data: Poisson Regression**
+1. **Robust Bayesian Regression**
+1. **Hierarchical Models**
+1. **Markov Chain Monte Carlo (MCMC) and Model Metrics**
+1. **Model Comparison: Cross-Validation and Other Metrics**
 
 ## Probabilistic Programming Languages (PPLs)
 
 Along with slides for the content, this repository also holds `Stan` code and also `Turing.jl` code for all models.
 `Stan` and `Turing.jl` represents, respectively, the present and future of [probabilistic programming](https://en.wikipedia.org/wiki/Probabilistic_programming) languages.
 
+All model files are tested in [GitHub Actions](https://github.com/storopoli/Bayesian-Statistics/actions/workflows/models.yml)
+against the latest `Stan` and Julia/`Turing.jl` versions.
+
 ### `Stan`
 
 [**`Stan`**](https://mc-stan.org) (Carpenter et al., 2017) `Stan` is a state-of-the-art platform for statistical modeling and high-performance statistical computation.

diff --git a/flake.nix b/flake.nix
@@ -42,6 +42,7 @@
           shellHook = ''
             export JULIA_NUM_THREADS="auto"
             export JULIA_PROJECT="turing"
+            export CMDSTAN_HOME="${pkgs.cmdstan}/opt/cmdstan"
             ${self.checks.${system}.pre-commit-check.shellHook}
           '';
         };

diff --git a/stan/02-linear_regression-QR.stan b/stan/02-linear_regression-QR.stan
@@ -41,9 +41,7 @@ generated quantities {
   vector[K] beta;
   beta = R_ast_inverse * theta; // coefficients on X scale
 }
-// results:All 4 chains finished successfully.
-Mean chain execution time: 0.2 seconds.
-Total execution time: 0.2 seconds.
+// results:
 //All 4 chains finished successfully.
 //Mean chain execution time: 0.2 seconds.
 //Total execution time: 0.2 seconds.

diff --git a/stan/03-logistic_regression-optimized.stan b/stan/03-logistic_regression-optimized.stan
@@ -24,7 +24,7 @@ model {
 
   // likelihood
   y ~ bernoulli_logit_glm(X, alpha, beta);
-  // no binomial_logit_glm available in Stan <=2.29
+  // no binomial_logit_glm available in Stan 2.33
 }
 // results:
 //All 4 chains finished successfully.

diff --git a/stan/08-robust_robit_regression.stan b/stan/08-robust_robit_regression.stan
@@ -26,7 +26,7 @@ model {
 
   // likelihood
   vector[N] p_hat;
-  for (n in 1:N) p_hat[n] = student_t_cdf(alpha + X[n] * beta, nu, 0, sqrt((nu - 2)/nu));
+  for (n in 1:N) p_hat[n] = student_t_cdf(alpha + X[n] * beta | nu, 0, sqrt((nu - 2)/nu));
   y ~ bernoulli(p_hat);
 }
 // results:

diff --git a/stan/10-robust_zero_inflated_regression-poisson-optimized.stan b/stan/10-robust_zero_inflated_regression-poisson-optimized.stan
@@ -29,10 +29,10 @@ model {
     if (y[n] == 0) {
       target += log_sum_exp(bernoulli_lpmf(1 | gamma),
                             bernoulli_lpmf(0 | gamma) +
-                            poisson_log_glm_lpmf(y[n] | X[n], alpha, beta));
+                            poisson_log_glm_lpmf(y[n] | to_matrix(X[n]), alpha, beta));
     } else {
       target += bernoulli_lpmf(0 | gamma) +
-                poisson_log_glm_lpmf(y[n] | X[n], alpha, beta);
+                poisson_log_glm_lpmf(y[n] | to_matrix(X[n]), alpha, beta);
     }
   }
 }

diff --git a/stan/11-hierarchical_varying_intercept-non_centered.stan b/stan/11-hierarchical_varying_intercept-non_centered.stan
@@ -28,7 +28,7 @@ parameters {
   vector[J] z;                           // group-level non-centered intercepts
 }
 transformed parameters {
-  vector[J] alpha_j = z * tau .+ alpha;           // group-level intercepts
+  vector[J] alpha_j = z * tau + alpha;           // group-level intercepts
 }
 model {
   // priors