diff --git a/python/prophet/forecaster.py b/python/prophet/forecaster.py
index 495ff7d..c3027e4 100644
--- a/python/prophet/forecaster.py
+++ b/python/prophet/forecaster.py
@@ -1442,7 +1442,7 @@ class Prophet(object):
             sim_values[k] = np.column_stack(v)
         return sim_values
 
-    def sample_model(self, df, seasonal_features, iteration, s_a, s_m):
+    def sample_model(self, df, seasonal_features, iteration, s_a, s_m) -> Dict[str, np.ndarray]:
         """Simulate observations from the extrapolated generative model.
 
         Parameters
@@ -1455,7 +1455,7 @@ class Prophet(object):
 
         Returns
         -------
-        Dataframe with trend and yhat, each like df['t'].
+        Dictionary with `yhat` and `trend`, each like df['t'].
         """
         trend = self.sample_predictive_trend(df, iteration)
 
@@ -1467,10 +1467,10 @@ class Prophet(object):
         sigma = self.params['sigma_obs'][iteration]
         noise = np.random.normal(0, sigma, df.shape[0]) * self.y_scale
 
-        return pd.DataFrame({
+        return {
             'yhat': trend * (1 + Xb_m) + Xb_a + noise,
             'trend': trend
-        })
+        }
 
     def sample_model_vectorized(
         self,
@@ -1480,12 +1480,12 @@ class Prophet(object):
         s_a: np.ndarray,
         s_m: np.ndarray,
         n_samples: int,
-    ) -> List[pd.DataFrame]:
+    ) -> List[Dict[str, np.ndarray]]:
         """Simulate observations from the extrapolated generative model. Vectorized version of sample_model().
 
         Returns
         -------
-        List (length n_samples) of DataFrames with np.arrays for trend and yhat, each ordered like df['t'].
+        List (length n_samples) of dictionaries with arrays for trend and yhat, each ordered like df['t'].
         """
         # Get the seasonality and regressor components, which are deterministic per iteration
         beta = self.params['beta'][iteration]
@@ -1499,10 +1499,10 @@ class Prophet(object):
 
         simulations = []
         for trend, noise in zip(trends, noise_terms):
-            simulations.append(pd.DataFrame({
+            simulations.append({
                 'yhat': trend * (1 + Xb_m) + Xb_a + noise,
                 'trend': trend
-            }))
+            })
         return simulations
 
     def sample_predictive_trend(self, df, iteration):