Having no accepted requests no longer crashes analytics (#257)

* Having no accepted requests no longer crashes analytics

* Modified groupby calls according to pandas FutureWarning

* Having no accepted requests no longer crashes analytics, part II

src/ridepy/util/analytics/requests.py

@@ -129,7 +129,7 @@ def _create_transportation_requests_dataframe(
     # If transportation as submitted were submitted, calculate more properties.
     # NOTE that these properties might equally well be computed using the
     # inferred requests, but in case of differences between the requests
-    # the resulting change in behavior might not intended. Therefore so far
+    # the resulting change in behavior might not be intended. Therefore, so far
     # we only compute these quantities if transportation_requests are submitted.
 
     # - direct travel time

src/ridepy/util/analytics/stops.py

@@ -161,7 +161,15 @@ def _add_locations_to_stoplist_dataframe(*, reqs, stops, space) -> pd.DataFrame:
 
     # use the requests' locations and reshape them into a DateFrame indexed by
     # `request_id` and `delta_occupancy`
-    locations = reqs.loc[:, ("accepted", ["origin", "destination"])]
+    if "accepted" in reqs.columns:
+        locations = reqs.loc[:, ("accepted", ["origin", "destination"])]
+    else:
+        locations = pd.DataFrame(
+            index=reqs.index,
+            columns=pd.MultiIndex.from_product(
+                [["accepted"], ["origin", "destination"]]
+            ),
+        )
     locations.columns = locations.columns.droplevel(0).rename("delta_occupancy")
     locations = locations.stack().rename("location")
 

src/ridepy/util/analytics/vehicle.py

@@ -1,3 +1,4 @@
+import numpy as np
 import pandas as pd
 
 
@@ -51,24 +52,32 @@ def get_vehicle_quantities(stops: pd.DataFrame, requests: pd.DataFrame) -> pd.Da
     total_dist_driven = stops.groupby("vehicle_id")["dist_to_next"].sum()
     total_time_driven = stops.groupby("vehicle_id")["time_to_next"].sum()
 
-    avg_direct_dist = serviced_requests.groupby(("serviced", "vehicle_id")).apply(
-        lambda gdf: gdf.submitted.direct_travel_distance.mean()
-    )
-
-    avg_direct_time = serviced_requests.groupby(("serviced", "vehicle_id")).apply(
-        lambda gdf: gdf.submitted.direct_travel_time.mean()
-    )
-
-    total_direct_dist = serviced_requests.groupby(("serviced", "vehicle_id")).apply(
-        lambda gdf: gdf.submitted.direct_travel_distance.sum()
-    )
-
-    total_direct_time = serviced_requests.groupby(("serviced", "vehicle_id")).apply(
-        lambda gdf: gdf.submitted.direct_travel_time.sum()
-    )
-
-    efficiency_dist = total_direct_dist / total_dist_driven
-    efficiency_time = total_direct_time / total_time_driven
+    if not serviced_requests.empty:
+        avg_direct_dist = serviced_requests.groupby(
+            ("serviced", "vehicle_id"), group_keys=False
+        ).apply(lambda gdf: gdf.submitted.direct_travel_distance.mean())
+
+        avg_direct_time = serviced_requests.groupby(
+            ("serviced", "vehicle_id"), group_keys=False
+        ).apply(lambda gdf: gdf.submitted.direct_travel_time.mean())
+
+        total_direct_dist = serviced_requests.groupby(
+            ("serviced", "vehicle_id"), group_keys=False
+        ).apply(lambda gdf: gdf.submitted.direct_travel_distance.sum())
+
+        total_direct_time = serviced_requests.groupby(
+            ("serviced", "vehicle_id"), group_keys=False
+        ).apply(lambda gdf: gdf.submitted.direct_travel_time.sum())
+
+        efficiency_dist = total_direct_dist / total_dist_driven
+        efficiency_time = total_direct_time / total_time_driven
+    else:
+        avg_direct_dist = pd.Series(np.nan, index=avg_occupancy.index, dtype=float)
+        avg_direct_time = pd.Series(np.nan, index=avg_occupancy.index, dtype=float)
+        total_direct_dist = pd.Series(np.nan, index=avg_occupancy.index, dtype=float)
+        total_direct_time = pd.Series(np.nan, index=avg_occupancy.index, dtype=float)
+        efficiency_dist = pd.Series(np.nan, index=avg_occupancy.index, dtype=float)
+        efficiency_time = pd.Series(np.nan, index=avg_occupancy.index, dtype=float)
 
     res = pd.DataFrame(
         dict(

test/test_analytics.py

@@ -939,3 +939,26 @@ def test_get_stops_and_requests_with_actual_simulation():
 
     assert len(stops) == 2020
     assert len(requests) == 1000
+
+
+def test_get_stops_and_requests_with_actual_simulation_none_accepted():
+    space = Euclidean1D()
+    rg = RandomRequestGenerator(rate=10, space=space)
+    transportation_requests = list(it.islice(rg, 1000))
+
+    fs = SlowSimpleFleetState(
+        initial_locations={k: 0 for k in range(10)},
+        seat_capacities=0,
+        space=space,
+        dispatcher=BruteForceTotalTravelTimeMinimizingDispatcher(),
+        vehicle_state_class=VehicleState,
+    )
+
+    events = list(fs.simulate(transportation_requests))
+
+    stops, requests = get_stops_and_requests(events=events, space=space)
+
+    assert len(stops) == 10 * 2  # only initial and final stops
+    assert len(requests) == 1000
+
+    vehicle_quantities = get_vehicle_quantities(stops, requests)