Merge pull request #331 from openclimatefix/issue/302-assign-pv-row-number

add pv_system_row_number

Author: peterdudfield

nowcasting_dataset/data_sources/fake.py

@@ -239,6 +239,10 @@ def create_gsp_pv_dataset(
     data["x_coords"] = x_coords
     data["y_coords"] = y_coords
 
+    # Add 1000 to the id numbers for the row numbers.
+    # This is a quick way to make sure row number is different from id,
+    data["pv_system_row_number"] = data["id"] + 1000
+
     data.__setitem__("data", data.data.clip(min=0))
 
     return data

nowcasting_dataset/data_sources/pv/pv_data_source.py

@@ -228,11 +228,7 @@ def get_example(
 
         selected_pv_power = selected_pv_power[all_pv_system_ids]
 
-        # TODO: Issue #302. pv_system_row_number is assigned to but never used.
-        # That may indicate a bug?
-        pv_system_row_number = np.flatnonzero(  # noqa: F841
-            self.pv_metadata.index.isin(all_pv_system_ids)
-        )
+        pv_system_row_number = np.flatnonzero(self.pv_metadata.index.isin(all_pv_system_ids))
         pv_system_x_coords = self.pv_metadata.location_x[all_pv_system_ids]
         pv_system_y_coords = self.pv_metadata.location_y[all_pv_system_ids]
 
@@ -267,8 +263,16 @@ def get_example(
                 id_index=range(len(all_pv_system_ids.values)),
             ),
         )
+        pv_system_row_number = xr.DataArray(
+            data=pv_system_row_number,
+            dims=["id_index"],
+            coords=dict(
+                id_index=range(len(all_pv_system_ids.values)),
+            ),
+        )
         pv["x_coords"] = x_coords
         pv["y_coords"] = y_coords
+        pv["pv_system_row_number"] = pv_system_row_number
 
         # pad out so that there are always n_pv_systems_per_example, pad with zeros
         pad_n = self.n_pv_systems_per_example - len(pv.id_index)

nowcasting_dataset/data_sources/pv/pv_model.py

@@ -16,4 +16,12 @@ def model_validation(cls, v):
         assert (~isnan(v.data)).all(), "Some pv data values are NaNs"
         assert (~isinf(v.data)).all(), "Some pv data values are Infinite"
         assert (v.data >= 0).all(), "Some pv data values are below 0"
+
+        assert v.time is not None
+        assert v.x_coords is not None
+        assert v.y_coords is not None
+        assert v.pv_system_row_number is not None
+
+        assert len(v.pv_system_row_number.shape) == 2
+
         return v