Validation set for Gradient Based Models woohoo

cca_zoo/probabilistic/_cca.py

@@ -54,7 +54,7 @@ def __init__(
         copy_data=True,
         random_state: int = 0,
         learning_rate=1e-1,
-        n_iter=40000,
+        n_iter=20000,
         num_samples=5000,
         num_warmup=5000,
     ):
@@ -214,31 +214,6 @@ def _guide(self, views):
         with numpyro.plate("n", self.n_samples_):
             z = numpyro.sample("z", dist.MultivariateNormal(z_loc, jnp.diag(z_scale)))
 
-    def transform(self, views: Iterable[np.ndarray], y=None, return_std=False):
-        """
-        Transform the data into the latent space.
-
-        Parameters
-        ----------
-        views : Iterable[np.ndarray]
-            A list or tuple of numpy arrays representing different representations of the same samples. Each numpy array must have the same number of rows.
-        y: Any, optional
-            Ignored in this implementation.
-
-        Returns
-        -------
-        representations : np.ndarray
-            The transformed data in the latent space.
-        """
-        svi = SVI(
-            self._model,
-            self._guide,
-            numpyro.optim.Adam(self.learning_rate),
-            loss=numpyro.infer.Trace_ELBO(),
-        )
-        svi_result = svi.run(self.rng_key, self.n_iter, views)
-        return np.array(svi_result.params["z"])
-
     def render(self, views):
         # check if graphviz is installed
         try:

test/test_explained_variance.py

@@ -54,7 +54,7 @@ def test_transformed_covariance_ratio(toy_model, synthetic_views):
     pls_cov_ratios = pls.explained_covariance_ratio(synthetic_views)
     # sum of these should be 1 within a small tolerance
     assert np.isclose(
-        np.sum(pls_cov_ratios), 1, atol=1e-2
+        np.sum(pls_cov_ratios), 1, atol=2e-2
     ), "Expected sum of ratios to be 1"
 
     cov_ratios = toy_model.explained_covariance_ratio(synthetic_views)

test/test_probabilistic.py

@@ -2,8 +2,9 @@
 import pytest
 
 from cca_zoo.datasets import LatentVariableData
-from cca_zoo.linear import CCA
+from cca_zoo.linear import CCA, PLS
 from cca_zoo.probabilistic import ProbabilisticCCA
+from cca_zoo.probabilistic._pls import ProbabilisticPLS
 
 
 @pytest.fixture
@@ -32,13 +33,24 @@ def test_cca_vs_probabilisticCCA(setup_data):
 
     # Assert: Calculate correlation coefficient and ensure it's greater than 0.95
     z = cca.transform([X, Y])[0]
-    z_p = np.array(pcca.transform([X, Y]))
     # correlation between cca and pcca
-    correlation_matrix = np.abs(np.corrcoef(z.reshape(-1), z_p.reshape(-1)))
+    correlation_matrix = np.abs(np.corrcoef(z.reshape(-1), pcca.params["z_loc"].reshape(-1)))
     correlation = correlation_matrix[0, 1]
 
+    assert correlation > 0.8
 
-#
-#     assert (
-#         correlation > 0.9
-#     ), f"Expected correlation greater than 0.95, got {correlation}"
+def test_pls_vs_probabilisticPLS(setup_data):
+    X, Y, data = setup_data
+    # Models and fit
+    pls = PLS(latent_dimensions=1)
+    ppls = ProbabilisticPLS(latent_dimensions=1, random_state=10)
+    pls.fit([X, Y])
+    ppls.fit([X, Y])
+
+    # Assert: Calculate correlation coefficient and ensure it's greater than 0.95
+    z = pls.transform([X, Y])[0]
+    # correlation between cca and pcca
+    correlation_matrix = np.abs(np.corrcoef(z.reshape(-1), ppls.params["z_loc"].reshape(-1)))
+    correlation = correlation_matrix[0, 1]
+
+    assert correlation > 0.8

test/test_sequential.py

@@ -10,7 +10,7 @@
 # Fixtures
 @pytest.fixture
 def simulated_data():
-    data_generator = JointData(view_features=[4, 6], latent_dims=5, correlation=0.8)
+    data_generator = JointData(view_features=[4, 6], latent_dims=3, correlation=0.8)
     X, Y = data_generator.sample(50)
     return X, Y
 
@@ -32,7 +32,7 @@ def test_sequential_model_fits_and_identifies_effects(
     X, Y = simulated_data
     sequential_model = SequentialModel(
         grid_search_estimator,
-        latent_dimensions=10,
+        latent_dimensions=4,
         permutation_test=True,
         p_threshold=0.05,
     )