def test_all_player_features(self):
season = Season(2013)
all_players = players.find({}).sort('id', pymongo.ASCENDING)
offensive_features = []
defensive_features = []
for player_data in all_players:
player = Player(player_data['id'])
print 'Extracting offensive features for {} from the {}'.format(player, season)
o_features = player_ocluster_features(player, season)
print 'Extracting defensive features for {} from the {}'.format(player, season)
d_features = player_dcluster_features(player, season)
offensive_features.append(o_features)
defensive_features.append(d_features)
offensive_features = np.array(offensive_features)
defensive_features = np.array(defensive_features)
o_features_file = os.path.join('season_data', '2013', 'player_offense_features.csv')
d_features_file = os.path.join('season_data', '2013', 'player_defense_features.csv')
o_header = 'id,ast_pct,ts_pct,orb_pct,usg,ortg,mp_pct'
d_header = 'id,blk_pct,stl_pct,drb_pct,drtg,pf_pct,mp_pct'
np.savetxt(o_features_file, offensive_features, delimiter=',', header=o_header)
np.savetxt(d_features_file, defensive_features, delimiter=',', header=d_header)
self.assertTrue(os.path.exists(o_features_file))
self.assertTrue(os.path.exists(d_features_file))
def test_all_player_features(self):
season = Season(2013)
all_players = players.find({}).sort('id', pymongo.ASCENDING)
offensive_features = []
for player_data in all_players:
player = Player(player_data['id'])
print 'Extracting offensive features for {} from the {}'.format(player, season)
features = player_ocluster_features(player, season)
offensive_features.append(features)
offensive_features = np.array(offensive_features)
def extract_data(filename='shots.csv'):
r, c, s, g, t, p = dict(), dict(), dict(), dict(), dict(), dict()
for region in regions.find():
player_id = region['regionId']
r[player_id] = region['name']
for tournament in tournaments.find():
player_id = tournament['tournamentId']
c[player_id] = tournament['name']
for season in seasons.find():
player_id = season['seasonId']
s[player_id] = season['name']
for stage in stages.find():
player_id = stage['stageId']
g[player_id] = stage['name']
for team in teams.find():
player_id = team['teamId']
t[player_id] = team['name']
for player in players.find():
player_id = player['playerId']
p[player_id] = player['name']
f = open(filename, 'w', newline='\n')
writer = csv.DictWriter(f, fieldnames=FIELDNAMES + QUALIFIERS, extrasaction='ignore')
writer.writeheader()
# goals = events.count({'isGoal': True, 'isOwnGoal': {'$exists': False}})
# shots = events.count({'isShot': True, 'isOwnGoal': {'$exists': False}})
# print('{0:,} goals from {1:,} shots ({2:.1%} shot rate)'.format(goals, shots, goals / shots))
for event in events.find({'isShot': True, 'isOwnGoal': {'$exists': False}}).sort('matchId', -1):
event['isGoal'] = 1 if event.get('isGoal') else 0
event['Region'] = r[event['regionId']] if event.get('regionId') else None
event['Tournament'] = c[event['tournamentId']] if event.get('tournamentId') else None
event['Season'] = s[event['seasonId']] if event.get('seasonId') else None
event['Stage'] = g[event['stageId']] if event.get('stageId') else None
event['Team'] = t[event['teamId']] if event.get('teamId') else None
try:
event['Player'] = p[event['playerId']] if event.get('playerId') else None
except KeyError:
player_id = event['playerId']
print('Missing playerId: {}'.format(player_id))
get_player(player_id)
event['Player'] = None
event_qualifiers = {q['type']['displayName']: q.get('value', 1) for q in event['qualifiers']}
for qualifier in QUALIFIERS:
event[qualifier] = event_qualifiers.get(qualifier, 0)
writer.writerow(event)
def calc_all_player_times(year, recompute=False):
class TimeComputationError(Exception):
def __init__(self, msg):
self.msg = msg
def __str__(self):
return self.msg
all_players = players.find({}).sort('id', pymongo.ASCENDING)
print('Loading the {}-{} NBA season'.format(year, year + 1))
season = Season(year)
print('Loaded season')
print('Computing time on court for all players in all games...')
for player_data in all_players:
player = Player(player_data['id'])
games_played = season.get_player_games_in_range(player)
for game in games_played:
print('Calculating time on court for {} ({}) in {} ({})'.format(player, player.id, game, game.id))
boxscore_minutes = game.player_boxscore(player)['totalSecondsPlayed'] / 60.0
if boxscore_minutes > 0:
time_on_court = player.time_on_court(game, recompute=recompute)
computed_minutes = compute_ts_length(time_on_court, unit='minutes')
else:
# there's never anything to calculate anyway
computed_minutes = 0
if not abs(computed_minutes - boxscore_minutes) <= 0.5:
print('In computing playing time for {} ({}) in {} ({}):'.format(player, player.id, game, game.id),
file=sys.stderr)
print('Discrepancy between computed time: {0:2.2f}, and boxscore time: {1:2.2f}'.format(computed_minutes, boxscore_minutes),
file=sys.stderr)
#raise TimeComputationError('Discrepancy between computed time: {}, and boxscore time: {}'.format(computed_minutes, boxscore_minutes)
else:
print('{} played {} minutes in {}'.format(player, round(computed_minutes, 3), game))
def get_all_player_features(season, start_date=None, end_date=None, recompute=False):
all_players = players.find({}).sort('id', pymongo.ASCENDING)
o_header = ['ast_pct', 'ts_pct', 'orb_pct', 'usg', 'ortg', 'mp_pct']
d_header = ['blk_pct', 'stl_pct', 'drb_pct', 'drtg', 'pf_pct']
str_format = '%Y-%m-%d'
feature_file = 'player-features-from-{}-to-{}.csv'.format(start_date.strftime(str_format), end_date.strftime(str_format))
path = os.path.join('season_data', str(season.season), feature_file)
if os.path.exists(path) and not recompute:
player_features = pd.read_csv(path, index_col=0)
else:
feature_headers = o_header + d_header
index = []
all_features = []
for i, player_data in enumerate(all_players):
player = Player(player_data['id'])
index.append(player.id)
#print 'Extracting offensive features for {} from the {}'.format(player, season)
o_features = player_ocluster_features(player, season, start_date, end_date)
#print 'Extracting defensive features for {} from the {}'.format(player, season)
d_features = player_dcluster_features(player, season, start_date, end_date)
# drop the id in d_features
all_features.append(o_features[1:] + d_features[1:-1])
player_features = pd.DataFrame(data=all_features, columns=feature_headers, index=index)
player_features.index.name = 'id'
player_features.to_csv(path)
return player_features
def extract_data(filename='shots.csv'):
r, c, s, g, t, p = dict(), dict(), dict(), dict(), dict(), dict()
for region in regions.find():
player_id = region['regionId']
r[player_id] = region['name']
for tournament in tournaments.find():
player_id = tournament['tournamentId']
c[player_id] = tournament['name']
for season in seasons.find():
player_id = season['seasonId']
s[player_id] = season['name']
for stage in stages.find():
player_id = stage['stageId']
g[player_id] = stage['name']
for team in teams.find():
player_id = team['teamId']
t[player_id] = team['name']
for player in players.find():
player_id = player['playerId']
p[player_id] = player['name']
f = open(filename, 'w', newline='\n')
writer = csv.DictWriter(f,
fieldnames=FIELDNAMES + QUALIFIERS,
extrasaction='ignore')
writer.writeheader()
# goals = events.count({'isGoal': True, 'isOwnGoal': {'$exists': False}})
# shots = events.count({'isShot': True, 'isOwnGoal': {'$exists': False}})
# print('{0:,} goals from {1:,} shots ({2:.1%} shot rate)'.format(goals, shots, goals / shots))
for event in events.find({
'isShot': True,
'isOwnGoal': {
'$exists': False
}
}).sort('matchId', -1):
event['isGoal'] = 1 if event.get('isGoal') else 0
event['Region'] = r[event['regionId']] if event.get(
'regionId') else None
event['Tournament'] = c[event['tournamentId']] if event.get(
'tournamentId') else None
event['Season'] = s[event['seasonId']] if event.get(
'seasonId') else None
event['Stage'] = g[event['stageId']] if event.get('stageId') else None
event['Team'] = t[event['teamId']] if event.get('teamId') else None
try:
event['Player'] = p[event['playerId']] if event.get(
'playerId') else None
except KeyError:
player_id = event['playerId']
print('Missing playerId: {}'.format(player_id))
get_player(player_id)
event['Player'] = None
event_qualifiers = {
q['type']['displayName']: q.get('value', 1)
for q in event['qualifiers']
}
for qualifier in QUALIFIERS:
event[qualifier] = event_qualifiers.get(qualifier, 0)
writer.writerow(event)