def test_convert_to_actions(self):
df_actions = sb.convert_to_actions(self.events_japbel, 782)
assert len(df_actions) > 0
SPADLSchema.validate(df_actions)
assert (df_actions.game_id == 7584).all()
assert ((df_actions.team_id == 782) |
(df_actions.team_id == 778)).all()
def convert_statsbomb_data():
seasons = {
3: '2018',
}
leagues = {
'FIFA World Cup': 'WorldCup',
}
spadl_datafolder = os.path.join(_data_dir, 'statsbomb')
free_open_data_remote = 'https://raw.githubusercontent.com/statsbomb/open-data/master/data/'
SBL = statsbomb.StatsBombLoader(root=free_open_data_remote, getter='remote')
# View all available competitions
df_competitions = SBL.competitions()
df_selected_competitions = df_competitions[
df_competitions.competition_name.isin(leagues.keys())
]
for competition in df_selected_competitions.itertuples():
# Get games from all selected competition
games = SBL.games(competition.competition_id, competition.season_id)
games_verbose = tqdm(list(games.itertuples()), desc='Loading match data')
teams, players = [], []
competition_id = leagues[competition.competition_name]
season_id = seasons[competition.season_id]
spadl_h5 = os.path.join(spadl_datafolder, f'spadl-{competition_id}-{season_id}.h5')
with pd.HDFStore(spadl_h5) as spadlstore:
spadlstore.put('actiontypes', spadl.actiontypes_df(), format='table')
spadlstore.put('results', spadl.results_df(), format='table')
spadlstore.put('bodyparts', spadl.bodyparts_df(), format='table')
for game in games_verbose:
# load data
teams.append(SBL.teams(game.game_id))
players.append(SBL.players(game.game_id))
events = SBL.events(game.game_id)
# convert data
spadlstore.put(
f'actions/game_{game.game_id}',
statsbomb.convert_to_actions(events, game.home_team_id),
format='table',
)
games.season_id = season_id
games.competition_id = competition_id
spadlstore.put('games', games)
spadlstore.put(
'teams',
pd.concat(teams).drop_duplicates('team_id').reset_index(drop=True),
)
spadlstore.put(
'players',
pd.concat(players).drop_duplicates('player_id').reset_index(drop=True),
)
def test_convert_own_goal(self):
events_morira = self.SBL.events(7577)
own_goal_for_event = events_morira[
events_morira.event_id == '8981bc58-6041-4b78-95c5-ebe9677ca379']
own_goal_for_actions = sb.convert_to_actions(own_goal_for_event, 797)
assert len(own_goal_for_actions) == 0
own_goal_against_event = events_morira[
events_morira.event_id == 'cef0fcb6-28d0-49d7-8f93-4a0aef28001a']
own_goal_against_actions = sb.convert_to_actions(
own_goal_against_event, 797)
assert len(own_goal_against_actions) == 1
assert own_goal_against_actions.iloc[0][
'type_id'] == spadl.actiontypes.index('bad_touch')
assert own_goal_against_actions.iloc[0][
'result_id'] == spadl.results.index('owngoal')
assert own_goal_against_actions.iloc[0][
'bodypart_id'] == spadl.bodyparts.index('foot')
def test_convert_pass(self):
pass_event = self.events_japbel[self.events_japbel.event_id ==
'a1b55211-a292-4294-887b-5385cc3c5705']
pass_action = sb.convert_to_actions(pass_event, self.id_bel).iloc[0]
assert pass_action['team_id'] == 782
assert pass_action['player_id'] == 3289
assert pass_action['type_id'] == spadl.actiontypes.index('pass')
assert pass_action['result_id'] == spadl.results.index('success')
assert pass_action['bodypart_id'] == spadl.bodyparts.index('foot')
def test_convert_time(self, period, timestamp, minute, second):
event = self.events_japbel[
self.events_japbel.event_id ==
'a1b55211-a292-4294-887b-5385cc3c5705'].copy()
event['period_id'] = period
event['timestamp'] = timestamp
event['minute'] = minute
event['second'] = second
action = sb.convert_to_actions(event, self.id_bel).iloc[0]
assert action['period_id'] == period
assert (action['time_seconds'] == 60 * minute -
((period > 1) * 45 * 60) - ((period > 2) * 45 * 60) -
((period > 3) * 15 * 60) - ((period > 4) * 15 * 60) + second)
matches = pd.concat(matches, sort=True).reset_index(drop=True)
# Load and convert match data
matches_verbose = tqdm.tqdm(list(matches.itertuples()),
desc="Loading match data")
teams, players, player_games = [], [], []
actions = {}
for match in matches_verbose:
teams.append(SBL.teams(match.match_id))
players.append(SBL.players(match.match_id))
events = SBL.events(match.match_id)
player_games.append(statsbomb.extract_player_games(events))
actions[match.match_id] = statsbomb.convert_to_actions(
events, match.home_team_id)
# Store converted spadl data in a h5-file
games = matches.rename(columns={"match_id": "game_id"})
teams = pd.concat(teams,
sort=True).drop_duplicates("team_id").reset_index(drop=True)
players = pd.concat(
players, sort=True).drop_duplicates("player_id").reset_index(drop=True)
player_games = pd.concat(player_games, sort=True).reset_index(drop=True)
spadl_h5 = os.path.join(datafolder, "spadl-statsbomb.h5")
with pd.HDFStore(spadl_h5) as spadlstore:
spadlstore["competitions"] = selected_competitions
spadlstore["games"] = games
spadlstore["teams"] = teams
spadlstore["players"] = players
def test_convert_end_location(self):
event = self.events_japbel[self.events_japbel.event_id ==
'a1b55211-a292-4294-887b-5385cc3c5705']
action = sb.convert_to_actions(event, self.id_bel).iloc[0]
assert action['end_x'] == ((49.0 - 1) / 119) * spadl.field_length
assert action['end_y'] == 68 - ((43.0 - 1) / 79) * spadl.field_width
def test_convert_start_location(self):
event = pd.DataFrame([{
'event_id':
'a1b55211-a292-4294-887b-5385cc3c5705',
'index':
5,
'period_id':
1,
'timestamp':
'00:00:00.920',
'minute':
0,
'second':
0,
'type_id':
30,
'type_name':
'Pass',
'possession':
2,
'possession_team_id':
782,
'possession_team_name':
'Belgium',
'play_pattern_id':
9,
'play_pattern_name':
'From Kick Off',
'team_id':
782,
'team_name':
'Belgium',
'duration':
0.973,
'extra': {
'pass': {
'recipient': {
'id': 5642,
'name': 'Axel Witsel'
},
'length': 12.369317,
'angle': 2.896614,
'height': {
'id': 1,
'name': 'Ground Pass'
},
'type': {
'id': 65,
'name': 'Kick Off'
},
'body_part': {
'id': 40,
'name': 'Right Foot'
},
}
},
'related_events': ['051449c5-e183-46f9-965d-1d8f00f017cb'],
'player_id':
3289.0,
'player_name':
'Romelu Lukaku Menama',
'position_id':
23.0,
'position_name':
'Center Forward',
'location': [61.0, 40.0],
'under_pressure':
None,
'counterpress':
None,
'game_id':
7584,
}])
action = sb.convert_to_actions(event, 782).iloc[0]
assert action['start_x'] == ((61.0 - 1) / 119) * spadl.field_length
assert action['start_y'] == 68 - ((40.0 - 1) / 79) * spadl.field_width
assert action['end_x'] == action['start_x']
assert action['end_y'] == action['start_y']
def test_convert_time(self, period, timestamp, minute, second):
event = pd.DataFrame([{
'event_id':
'a1b55211-a292-4294-887b-5385cc3c5705',
'index':
5,
'period_id':
period,
'timestamp':
timestamp,
'minute':
minute,
'second':
second,
'type_id':
30,
'type_name':
'Pass',
'possession':
2,
'possession_team_id':
782,
'possession_team_name':
'Belgium',
'play_pattern_id':
9,
'play_pattern_name':
'From Kick Off',
'team_id':
782,
'team_name':
'Belgium',
'duration':
0.973,
'extra': {
'pass': {
'recipient': {
'id': 5642,
'name': 'Axel Witsel'
},
'length': 12.369317,
'angle': 2.896614,
'height': {
'id': 1,
'name': 'Ground Pass'
},
'end_location': [49.0, 43.0],
'type': {
'id': 65,
'name': 'Kick Off'
},
'body_part': {
'id': 40,
'name': 'Right Foot'
},
}
},
'related_events': ['051449c5-e183-46f9-965d-1d8f00f017cb'],
'player_id':
3289.0,
'player_name':
'Romelu Lukaku Menama',
'position_id':
23.0,
'position_name':
'Center Forward',
'location': [61.0, 40.0],
'under_pressure':
None,
'counterpress':
None,
'game_id':
7584,
}])
action = sb.convert_to_actions(event, 782).iloc[0]
assert action['period_id'] == period
assert (action['time_seconds'] == 60 * minute -
((period > 1) * 45 * 60) - ((period > 2) * 45 * 60) -
((period > 3) * 15 * 60) - ((period > 4) * 15 * 60) + second)