Add own goals in StatsBomb deserializer

kloppy/domain/services/state_builder/builders/score.py

@@ -27,4 +27,9 @@ def reduce_after(self, state: Score, event: Event) -> Score:
                     state = replace(state, home=state.home + 1)
                 else:
                     state = replace(state, away=state.away + 1)
+            elif event.result == ShotResult.OWN_GOAL:
+                if event.team.ground == Ground.HOME:
+                    state = replace(state, away=state.away + 1)
+                else:
+                    state = replace(state, home=state.home + 1)
         return state

kloppy/infra/serializers/event/statsbomb/deserializer.py

@@ -53,6 +53,8 @@
 SB_EVENT_TYPE_CLEARANCE = 9
 SB_EVENT_TYPE_DRIBBLE = 14
 SB_EVENT_TYPE_SHOT = 16
+SB_EVENT_TYPE_OWN_GOAL_AGAINST = 20
+SB_EVENT_TYPE_OWN_GOAL_FOR = 25
 SB_EVENT_TYPE_GOALKEEPER_EVENT = 23
 SB_EVENT_TYPE_PASS = 30
 SB_EVENT_TYPE_50_50 = 33
@@ -856,6 +858,15 @@ def deserialize(self, inputs: StatsBombInputs) -> EventDataset:
                         **generic_event_kwargs,
                     )
                     new_events.append(shot_event)
+                elif event_type == SB_EVENT_TYPE_OWN_GOAL_AGAINST:
+                    shot_event = self.event_factory.build_shot(
+                        result=ShotResult.OWN_GOAL,
+                        qualifiers=[],
+                        **generic_event_kwargs,
+                    )
+                    new_events.append(shot_event)
+                elif event_type == SB_EVENT_TYPE_OWN_GOAL_FOR:
+                    pass
                 elif event_type == SB_EVENT_TYPE_CLEARANCE:
                     clearance_event_kwargs = _parse_clearance(
                         raw_event=raw_event, events=events
@@ -1063,7 +1074,7 @@ def deserialize(self, inputs: StatsBombInputs) -> EventDataset:
         # Last step is to add freeze_frame information
         for event in events:
             if event.event_type == EventType.SHOT:
-                if "freeze_frame" in event.raw_event["shot"]:
+                if "freeze_frame" in event.raw_event.get("shot", {}):
                     event.freeze_frame = transformer.transform_frame(
                         _parse_freeze_frame(
                             freeze_frame=event.raw_event["shot"][

kloppy/tests/files/statsbomb_event.json

@@ -170249,6 +170249,82 @@
     "aerial_won" : true
   }
 }, {
+    "id" : "f942c5b5-df4b-4ee4-9e90-ed5f5",
+    "index" : 4005,
+    "period" : 2,
+    "timestamp" : "00:44:51.456",
+    "minute" : 44,
+    "second" : 51,
+    "type" : {
+      "id" : 25,
+      "name" : "Own Goal For"
+    },
+    "possession" : 101,
+    "possession_team" : {
+      "id" : 206,
+      "name" : "Deportivo Alavés"
+    },
+    "play_pattern" : {
+      "id" : 2,
+      "name" : "From Corner"
+    },
+    "obv_for_after" : null,
+    "obv_for_before" : null,
+    "obv_for_net" : null,
+    "obv_against_after" : null,
+    "obv_against_before" : null,
+    "obv_against_net" : null,
+    "obv_total_net" : null,
+    "team" : {
+      "id" : 206,
+      "name" : "Deportivo Alavés"
+    },
+    "location" : [ 115.7, 39.9 ],
+    "duration" : 0.0,
+    "related_events" : [ "89dd4f4b-0a70-48d8-a0e7-ac4c" ]
+  }, {
+    "id" : "89dd4f4b-0a70-48d8-a0e7-ac4c",
+    "index" : 4006,
+    "period" : 2,
+    "timestamp" : "00:44:51.456",
+    "minute" : 44,
+    "second" : 51,
+    "type" : {
+      "id" : 20,
+      "name" : "Own Goal Against"
+    },
+    "possession" : 101,
+    "possession_team" : {
+      "id" : 206,
+      "name" : "Deportivo Alavés"
+    },
+    "play_pattern" : {
+      "id" : 2,
+      "name" : "From Corner"
+    },
+    "obv_for_after" : null,
+    "obv_for_before" : null,
+    "obv_for_net" : null,
+    "obv_against_after" : null,
+    "obv_against_before" : null,
+    "obv_against_net" : null,
+    "obv_total_net" : null,
+    "team" : {
+      "id" : 217,
+      "name" : "Barcelona"
+    },
+    "player" : {
+      "id" : 6629,
+      "name" : "Fernando Pacheco Flores"
+    },
+    "position" : {
+      "id" : 5,
+      "name" : "Left Center Back"
+    },
+    "location" : [ 4.4, 40.2 ],
+    "duration" : 0.0,
+    "related_events" : [ "f942c5b5-df4b-4ee4-9e90-ed5f5" ]
+  }, {
   "id" : "e1cc4d5e-ba55-4b6b-88cc-dae13311c1d9",
   "index" : 4001,
   "period" : 2,
@@ -170388,4 +170464,4 @@
     }
   }
 }
-]
+]

kloppy/tests/test_helpers.py

@@ -376,7 +376,7 @@ def test_event_dataset_to_polars(self, base_dir):
         import polars as pl
 
         c = df.select(pl.col("event_id").count())[0, 0]
-        assert c == 4041
+        assert c == 4042
 
     def test_tracking_dataset_to_polars(self):
         """

kloppy/tests/test_state_builder.py

@@ -32,7 +32,8 @@ def test_score_state_builder(self, base_dir):
             "0-0": 2909,
             "1-0": 717,
             "2-0": 405,
-            "3-0": 10,
+            "3-0": 7,
+            "3-1": 4,
         }
 
     def test_sequence_state_builder(self, base_dir):
@@ -93,7 +94,7 @@ def test_formation_state_builder(self, base_dir):
 
         # inspect FormationChangeEvent usage and formation state_builder
         assert events_per_formation_change["4-1-4-1"] == 3085
-        assert events_per_formation_change["4-4-2"] == 956
+        assert events_per_formation_change["4-4-2"] == 957
 
         assert dataset.metadata.teams[0].starting_formation == FormationType(
             "4-4-2"

kloppy/tests/test_statsbomb.py

@@ -15,8 +15,8 @@
     Point,
     Provider,
     FormationType,
-    Frame,
     Position,
+    ShotResult,
 )
 
 from kloppy import statsbomb
@@ -55,7 +55,7 @@ def test_correct_deserialization(
 
         assert dataset.metadata.provider == Provider.STATSBOMB
         assert dataset.dataset_type == DatasetType.EVENT
-        assert len(dataset.events) == 4041
+        assert len(dataset.events) == 4042
         assert len(dataset.metadata.periods) == 2
         assert (
             dataset.metadata.orientation == Orientation.ACTION_EXECUTING_TEAM
@@ -170,12 +170,12 @@ def test_correct_deserialization(
         )
 
         assert (
-            dataset.events[4039].get_qualifier_value(GoalkeeperQualifier)
+            dataset.events[4040].get_qualifier_value(GoalkeeperQualifier)
             == GoalkeeperActionType.SMOTHER
         )
 
         assert (
-            dataset.events[4040].get_qualifier_value(GoalkeeperQualifier)
+            dataset.events[4041].get_qualifier_value(GoalkeeperQualifier)
             == GoalkeeperActionType.PUNCH
         )
 
@@ -254,6 +254,36 @@ def test_foul_committed(self, lineup_data: Path, event_data: Path):
 
         assert len(dataset.events) == 23
 
+    def test_own_goal(self, lineup_data: Path, event_data: Path):
+        """
+        Test own goal events.
+
+        The StatsBomb "Own Goal For" (id = 25) and one "Own Goal Against" (id = 20) events
+        should be converted to a single shot event with ShotResult.OWN_GOAL.
+        """
+        dataset = statsbomb.load(
+            lineup_data=lineup_data,
+            event_data=event_data,
+        )
+
+        # The Own Goal For event should be removed
+        own_goal_for_event = [
+            event
+            for event in dataset.events
+            if event.event_id == "f942c5b5-df4b-4ee4-9e90-ed5f5"
+        ]
+        assert len(own_goal_for_event) == 0
+
+        # The Own Goal Against event should be converted to a shot event
+        own_goal_against_event = [
+            event
+            for event in dataset.events
+            if event.event_id == "89dd4f4b-0a70-48d8-a0e7-ac4c"
+        ]
+        assert len(own_goal_against_event) == 1
+        assert own_goal_against_event[0].event_type == EventType.SHOT
+        assert own_goal_against_event[0].result == ShotResult.OWN_GOAL
+
     def test_related_events(self, lineup_data: Path, event_data: Path):
         dataset = statsbomb.load(
             lineup_data=lineup_data, event_data=event_data

kloppy/tests/test_to_records.py

@@ -29,7 +29,7 @@ def dataset(self, event_data: Path, lineup_data: Path) -> EventDataset:
 
     def test_default_columns(self, dataset: EventDataset):
         records = dataset.to_records()
-        assert len(records) == 4041
+        assert len(records) == 4042
         assert list(records[0].keys()) == [
             "event_id",
             "event_type",