Make PyMC 5.13.0 the minimum version

pymc_marketing/clv/models/pareto_nbd.py

@@ -240,11 +240,11 @@
             "purchase_covariate": self.purchase_covariate_cols,
             "dropout_covariate": self.dropout_covariate_cols,
             "obs_var": ["recency", "frequency"],
+            "customer_id": self.data["customer_id"],
         }
-        mutable_coords = {"customer_id": self.data["customer_id"]}
-        with pm.Model(coords=coords, coords_mutable=mutable_coords) as self.model:
+        with pm.Model(coords=coords) as self.model:
             if self.purchase_covariate_cols:
-                purchase_data = pm.MutableData(
+                purchase_data = pm.Data(
                     "purchase_data",
                     self.data[self.purchase_covariate_cols],
                     dims=["customer_id", "purchase_covariate"],
@@ -273,7 +273,7 @@
 
             # churn priors
             if self.dropout_covariate_cols:
-                dropout_data = pm.MutableData(
+                dropout_data = pm.Data(
                     "dropout_data",
                     self.data[self.dropout_covariate_cols],
                     dims=["customer_id", "dropout_covariate"],

pymc_marketing/mmm/delayed_saturated_mmm.py

@@ -203,11 +203,9 @@ def _generate_and_preprocess_model_data(  # type: ignore
         date_data = X[self.date_column]
         channel_data = X[self.channel_columns]
 
-        self.coords_mutable: dict[str, Any] = {
-            "date": date_data,
-        }
         coords: dict[str, Any] = {
             "channel": self.channel_columns,
+            "date": date_data,
         }
 
         new_X_dict = {
@@ -349,20 +347,17 @@ def build_model(
         self._generate_and_preprocess_model_data(X, y)
         with pm.Model(
             coords=self.model_coords,
-            coords_mutable=self.coords_mutable,
         ) as self.model:
             channel_data_ = pm.Data(
                 name="channel_data",
                 value=self.preprocessed_data["X"][self.channel_columns],
                 dims=("date", "channel"),
-                mutable=True,
             )
 
             target_ = pm.Data(
                 name="target",
                 value=self.preprocessed_data["y"],
                 dims="date",
-                mutable=True,
             )
             if self.time_varying_intercept | self.time_varying_media:
                 time_index = pm.Data(
@@ -443,7 +438,6 @@ def build_model(
                     name="control_data",
                     value=self.preprocessed_data["X"][self.control_columns],
                     dims=("date", "control"),
-                    mutable=True,
                 )
 
                 control_contributions = pm.Deterministic(
@@ -461,7 +455,6 @@ def build_model(
                         self.date_column
                     ].dt.dayofyear.to_numpy(),
                     dims="date",
-                    mutable=True,
                 )
 
                 def create_deterministic(x: pt.TensorVariable) -> None:
@@ -546,7 +539,6 @@ def channel_contributions_forward_pass(
         """
         coords = {
             **self.model_coords,
-            **self.coords_mutable,
         }
         with pm.Model(coords=coords):
             pm.Deterministic(

pyproject.toml

@@ -31,7 +31,7 @@ dependencies = [
     "numpy>=1.17",
     "pandas",
     # NOTE: Used as minimum pymc version with ci.yml `OLDEST_PYMC_VERSION`
-    "pymc>=5.12.0,<5.16.0",
+    "pymc>=5.13.0,<5.16.0",
     "scikit-learn>=1.1.1",
     "seaborn>=0.12.2",
     "xarray",

tests/clv/test_distributions.py

@@ -261,7 +261,7 @@ def test_pareto_nbd_sample_prior(
             s = pm.Gamma(name="s", alpha=5, beta=1, size=s_size)
             beta = pm.Gamma(name="beta", alpha=5, beta=1, size=beta_size)
 
-            T = pm.MutableData(name="T", value=np.array(10))
+            T = pm.Data(name="T", value=np.array(10))
 
             ParetoNBD(
                 name="pareto_nbd",
@@ -436,7 +436,7 @@ def test_beta_geo_beta_binom_sample_prior(
             gamma = pm.Normal(name="gamma", mu=gamma_true, sigma=1e-4, size=gamma_size)
             delta = pm.Normal(name="delta", mu=delta_true, sigma=1e-4, size=delta_size)
 
-            T = pm.MutableData(name="T", value=np.array(T_true))
+            T = pm.Data(name="T", value=np.array(T_true))
 
             BetaGeoBetaBinom(
                 name="beta_geo_beta_binom",

tests/mmm/test_lift_test.py

@@ -325,15 +325,13 @@ def test_add_lift_measurements_before_new_data(
     channels = ["organic", "paid", "social"]
     coords = {
         "channel": channels,
-    }
-    coords_mutable = {
         "date": ["2020-01-01", "2020-01-02", "2020-01-03"],
     }
-    with pm.Model(coords=coords, coords_mutable=coords_mutable) as model:
+    with pm.Model(coords=coords) as model:
         alpha = pm.HalfNormal("alpha", dims="channel")
         lam = pm.HalfNormal("lam", dims="channel")
 
-        X = pm.Data("X", np.ones((3, 3)), dims=("date", "channel"), mutable=True)
+        X = pm.Data("X", np.ones((3, 3)), dims=("date", "channel"))
         pm.Deterministic(
             "random_operation",
             X + alpha + lam,

tests/test_model_builder.py

@@ -110,8 +110,8 @@ def build_model(self, X: pd.DataFrame, y: pd.Series, model_config=None):
         with pm.Model(coords=coords) as self.model:
             if model_config is None:
                 model_config = self.default_model_config
-            x = pm.MutableData("x", self.X["input"].values)
-            y_data = pm.MutableData("y_data", self.y)
+            x = pm.Data("x", self.X["input"].values)
+            y_data = pm.Data("y_data", self.y)
 
             # prior parameters
             a_loc = model_config["a"]["loc"]