MMM Case Study from PyData Global

docs/source/notebooks/index.md

@@ -16,6 +16,7 @@ mmm/mmm_time_varying_media_example
 mmm/mmm_components
 mmm/mmm_roas
 mmm/mmm_time_slice_cross_validation
+mmm/mmm_case_study
 :::
 
 :::{toctree}

pymc_marketing/mmm/mmm.py

@@ -2037,6 +2037,10 @@ def _create_synth_dataset(
     ) -> pd.DataFrame:
         """Create a synthetic dataset based on the given allocation strategy (Budget) and time granularity.
 
+        **Important**: When generating the posterior predicive distribution for the target with the optimized budget,
+        we are setting the control variables to zero! This is done because in many situations we do not have all the
+        control variables in the future (e.g. outlier control, special events).
+
         Parameters
         ----------
         df : pd.DataFrame
@@ -2131,6 +2135,7 @@ def allocate_budget_to_maximize_response(
         custom_constraints: dict[str, float] | None = None,
         quantile: float = 0.5,
         noise_level: float = 0.01,
+        **minimize_kwargs,
     ) -> az.InferenceData:
         """Allocate the given budget to maximize the response over a specified time period.
 
@@ -2144,6 +2149,10 @@ def allocate_budget_to_maximize_response(
         budget, and creates a synthetic dataset based on the optimal allocation. Finally,
         it performs posterior predictive sampling on the synthetic dataset.
 
+        **Important**: When generating the posterior predicive distribution for the target with the optimized budget,
+        we are setting the control variables to zero! This is done because in many situations we do not have all the
+        control variables in the future (e.g. outlier control, special events).
+
         Parameters
         ----------
         budget : float or int
@@ -2159,6 +2168,10 @@ def allocate_budget_to_maximize_response(
             Custom constraints for the optimization. If None, no custom constraints are applied.
         quantile : float, optional
             The quantile to use for recovering transformation parameters. Default is 0.5.
+        noise_level : float, optional
+            The level of noise added to the allocation strategy (by default 1%).
+        **minimize_kwargs
+            Additional arguments to pass to the `BudgetOptimizer`.
 
         Returns
         -------
@@ -2170,7 +2183,12 @@ def allocate_budget_to_maximize_response(
         ValueError
             If the time granularity is not supported.
 
+        ValueError
+            If the noise level is not a float.
         """
+        if not isinstance(noise_level, float):
+            raise ValueError("noise_level must be a float")
+
         parameters_mid = self.format_recovered_transformation_parameters(
             quantile=quantile
         )
@@ -2188,6 +2206,7 @@ def allocate_budget_to_maximize_response(
             total_budget=budget,
             budget_bounds=budget_bounds,
             custom_constraints=custom_constraints,
+            **minimize_kwargs,
         )
 
         synth_dataset = self._create_synth_dataset(

tests/mmm/test_mmm.py

@@ -503,6 +503,24 @@ def test_allocate_budget_to_maximize_response(self, mmm_fitted: MMM) -> None:
             inference_periods == num_periods
         ), f"Number of periods in the data {inference_periods} does not match the expected {num_periods}"
 
+    def test_allocate_budget_to_maximize_response_bad_noise_level(
+        self, mmm_fitted: MMM
+    ) -> None:
+        budget = 2.0
+        num_periods = 8
+        time_granularity = "weekly"
+        budget_bounds = {"channel_1": [0.5, 1.2], "channel_2": [0.5, 1.5]}
+        noise_level = "bad_noise_level"
+
+        with pytest.raises(ValueError, match="noise_level must be a float"):
+            mmm_fitted.allocate_budget_to_maximize_response(
+                budget=budget,
+                time_granularity=time_granularity,
+                num_periods=num_periods,
+                budget_bounds=budget_bounds,
+                noise_level=noise_level,
+            )
+
     @pytest.mark.parametrize(
         argnames="original_scale",
         argvalues=[False, True],