Fix: handling for context vars. add support for disabling prophet opt…

…ion. (#1209) * added proper handling of context variables * added unit test on prepare_data * update fix for prophet enabling * updated timestamp formats * update redundant dtype conversion * update notebook reference visuals * update notebook reference visuals * update notebook references * add back unit test file * clear nb outputs * remove nb previews * update pyproject toml file * add nb outputs * add nb outputs
facebookexperimental · Feb 5, 2025 · 025eded · 025eded
1 parent 98a2261
commit 025eded
Show file tree

Hide file tree

Showing 7 changed files with 1,106 additions and 812 deletions.
diff --git a/.gitignore b/.gitignore
@@ -1,6 +1,7 @@
 .DS_Store
 .Rproj.user
 .Rhistory
+.pyvenv/
 .venv/
 robynpy.egg-info/
 node_modules/

diff --git a/python/docs/robyn/modeling/feature_engineering.md b/python/docs/robyn/modeling/feature_engineering.md
diff --git a/python/pyproject.toml b/python/pyproject.toml
@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "robynpy"
-version = "0.1.1"
+version = "0.1.2"
 authors = [
   { name="Gufeng Zhou", email="[email protected]" },
   { name="Igor Skokan", email="[email protected]" },
@@ -36,6 +36,7 @@ dependencies = [
     "plotly",
     "nlopt",
     "cmake",
+    "ipykernel",
 ]
 [tool.setuptools.packages.find]
 where = ["src"]

diff --git a/python/src/robyn/modeling/feature_engineering.py b/python/src/robyn/modeling/feature_engineering.py
@@ -49,22 +49,34 @@ def perform_feature_engineering(self, quiet: bool = False) -> FeaturizedMMMData:
         dt_transform = self._prepare_data()
         self.logger.debug(f"Prepared data shape: {dt_transform.shape}")
 
-        if any(
-            var in self.holidays_data.prophet_vars
-            for var in ["trend", "season", "holiday", "monthly", "weekday"]
-        ):
+        # Check if Prophet decomposition should be performed
+        prophet_enabled = (
+            self.holidays_data is not None
+            and self.holidays_data.prophet_vars is not None
+            and len(self.holidays_data.prophet_vars) > 0
+        )
+
+        if prophet_enabled:
             self.logger.info("Starting Prophet decomposition")
             dt_transform = self._prophet_decomposition(dt_transform)
             if not quiet:
                 self.logger.info("Prophet decomposition complete")
+        else:
+            self.logger.info(
+                "Prophet decomposition disabled - no prophet variables specified"
+            )
 
         # Include all independent variables
-        all_ind_vars = (
-            self.holidays_data.prophet_vars
-            + self.mmm_data.mmmdata_spec.context_vars
+        all_ind_vars = []
+        if prophet_enabled:
+            all_ind_vars.extend(self.holidays_data.prophet_vars)
+
+        all_ind_vars.extend(
+            self.mmm_data.mmmdata_spec.context_vars
             + self.mmm_data.mmmdata_spec.paid_media_spends
             + self.mmm_data.mmmdata_spec.organic_vars
         )
+
         self.logger.debug(f"Processing {len(all_ind_vars)} independent variables")
 
         dt_mod = dt_transform
@@ -107,11 +119,48 @@ def _prepare_data(self) -> pd.DataFrame:
         dt_transform = self.mmm_data.data.copy()
         dt_transform["ds"] = pd.to_datetime(
             dt_transform[self.mmm_data.mmmdata_spec.date_var]
-        ).dt.strftime("%Y-%m-%d")
-        dt_transform["dep_var"] = dt_transform[self.mmm_data.mmmdata_spec.dep_var]
-        dt_transform["competitor_sales_B"] = dt_transform["competitor_sales_B"].astype(
-            "int64"
         )
+        dt_transform["dep_var"] = dt_transform[self.mmm_data.mmmdata_spec.dep_var]
+
+        # Set default factor_vars if None
+        if self.mmm_data.mmmdata_spec.factor_vars is None:
+            self.mmm_data.set_default_factor_vars()
+        factor_vars = self.mmm_data.mmmdata_spec.factor_vars or []
+
+        # Handle factor variables conversion first
+        for factor_var in factor_vars:
+            try:
+                dt_transform[factor_var] = dt_transform[factor_var].astype("category")
+                self.logger.debug(f"Converted {factor_var} to categorical")
+            except Exception as e:
+                self.logger.warning(
+                    f"Could not convert {factor_var} to categorical: {str(e)}"
+                )
+
+        # Only convert context variables that are used in numerical calculations
+        # i.e., those that aren't categorical/factor variables
+        numeric_context_vars = [
+            var
+            for var in self.mmm_data.mmmdata_spec.context_vars
+            if var not in factor_vars
+            and pd.api.types.is_object_dtype(dt_transform[var])
+        ]
+
+        if numeric_context_vars:
+            self.logger.debug(
+                f"Converting numeric context variables: {numeric_context_vars}"
+            )
+            for var in numeric_context_vars:
+                try:
+                    dt_transform[var] = pd.to_numeric(
+                        dt_transform[var], errors="coerce"
+                    )
+                    self.logger.debug(
+                        f"Converted {var} to numeric: {dt_transform[var].dtype}"
+                    )
+                except Exception as e:
+                    self.logger.warning(f"Could not convert {var} to numeric: {str(e)}")
+
         self.logger.debug("Data preparation complete")
         return dt_transform