MAINT Add parameter validation to LatentDirichletAllocation (scikit-learn#24212)

Co-authored-by: Thomas J. Fan <[email protected]>

Author: stefmolin

sklearn/decomposition/_lda.py

@@ -10,6 +10,7 @@
 
 # Author: Chyi-Kwei Yau
 # Author: Matthew D. Hoffman (original onlineldavb implementation)
+from numbers import Integral, Real
 
 import numpy as np
 import scipy.sparse as sp
@@ -21,6 +22,7 @@
 from ..utils.validation import check_non_negative
 from ..utils.validation import check_is_fitted
 from ..utils.fixes import delayed
+from ..utils._param_validation import Interval, StrOptions
 
 from ._online_lda_fast import (
     mean_change,
@@ -320,6 +322,25 @@ class conditional densities to the data and using Bayes' rule.
            [0.15297572, 0.00362644, 0.44412786, 0.39568399, 0.003586  ]])
     """
 
+    _parameter_constraints: dict = {
+        "n_components": [Interval(Integral, 0, None, closed="neither")],
+        "doc_topic_prior": [None, Interval(Real, 0, 1, closed="both")],
+        "topic_word_prior": [None, Interval(Real, 0, 1, closed="both")],
+        "learning_method": [StrOptions({"batch", "online"})],
+        "learning_decay": [Interval(Real, 0, 1, closed="both")],
+        "learning_offset": [Interval(Real, 1.0, None, closed="left")],
+        "max_iter": [Interval(Integral, 0, None, closed="left")],
+        "batch_size": [Interval(Integral, 0, None, closed="neither")],
+        "evaluate_every": [Interval(Integral, None, None, closed="neither")],
+        "total_samples": [Interval(Real, 0, None, closed="neither")],
+        "perp_tol": [Interval(Real, 0, None, closed="left")],
+        "mean_change_tol": [Interval(Real, 0, None, closed="left")],
+        "max_doc_update_iter": [Interval(Integral, 0, None, closed="left")],
+        "n_jobs": [None, Integral],
+        "verbose": ["verbose"],
+        "random_state": ["random_state"],
+    }
+
     def __init__(
         self,
         n_components=10,
@@ -357,26 +378,6 @@ def __init__(
         self.verbose = verbose
         self.random_state = random_state
 
-    def _check_params(self):
-        """Check model parameters."""
-        if self.n_components <= 0:
-            raise ValueError("Invalid 'n_components' parameter: %r" % self.n_components)
-
-        if self.total_samples <= 0:
-            raise ValueError(
-                "Invalid 'total_samples' parameter: %r" % self.total_samples
-            )
-
-        if self.learning_offset < 0:
-            raise ValueError(
-                "Invalid 'learning_offset' parameter: %r" % self.learning_offset
-            )
-
-        if self.learning_method not in ("batch", "online"):
-            raise ValueError(
-                "Invalid 'learning_method' parameter: %r" % self.learning_method
-            )
-
     def _init_latent_vars(self, n_features):
         """Initialize latent variables."""
 
@@ -559,7 +560,7 @@ def partial_fit(self, X, y=None):
         self
             Partially fitted estimator.
         """
-        self._check_params()
+        self._validate_params()
         first_time = not hasattr(self, "components_")
         X = self._check_non_neg_array(
             X, reset_n_features=first_time, whom="LatentDirichletAllocation.partial_fit"
@@ -609,7 +610,7 @@ def fit(self, X, y=None):
         self
             Fitted estimator.
         """
-        self._check_params()
+        self._validate_params()
         X = self._check_non_neg_array(
             X, reset_n_features=True, whom="LatentDirichletAllocation.fit"
         )

sklearn/decomposition/tests/test_online_lda.py

@@ -152,22 +152,6 @@ def test_lda_fit_transform(method):
     assert_array_almost_equal(X_fit, X_trans, 4)
 
 
-def test_invalid_params():
-    # test `_check_params` method
-    X = np.ones((5, 10))
-
-    invalid_models = (
-        ("n_components", LatentDirichletAllocation(n_components=0)),
-        ("learning_method", LatentDirichletAllocation(learning_method="unknown")),
-        ("total_samples", LatentDirichletAllocation(total_samples=0)),
-        ("learning_offset", LatentDirichletAllocation(learning_offset=-1)),
-    )
-    for param, model in invalid_models:
-        regex = r"^Invalid %r parameter" % param
-        with pytest.raises(ValueError, match=regex):
-            model.fit(X)
-
-
 def test_lda_negative_input():
     # test pass dense matrix with sparse negative input.
     X = np.full((5, 10), -1.0)

sklearn/tests/test_common.py

@@ -467,7 +467,6 @@ def test_estimators_do_not_raise_errors_in_init_or_set_params(Estimator):
     "ClassifierChain",
     "DictionaryLearning",
     "HashingVectorizer",
-    "LatentDirichletAllocation",
     "MiniBatchDictionaryLearning",
     "MultiTaskElasticNet",
     "MultiTaskLasso",