def generate_game_data(filename, teamno, folderpath=PLAY_BY_PLAY_DIRECTORY):
'''This funciton generates game data from lineup retrieved from stats.nba.com'''
df = pd.read_csv(folderpath + filename)
# add_aid_rows(df,)
teamname = filename[9:12] if teamno == 1 else filename[12:15]
team = BasketballGame(filename, df, teamname, teamno)
gamestats = []
# reads in data of the season's mean
ref = pd.read_csv(STATS_NBA_PATH % "2008-09_team_mean.csv") # reads the mean data of teams of the season
opponent_team_mean = ref[ref['TEAM_ABBREVIATION'] == team.oppteamname].squeeze()
opponent_team_mean.index = ['OPP_%s' % elt for elt in opponent_team_mean.index]
for lineup_no in range(1, team.totalcombi + 1): # Loop over lineups in the game
lineup = team.give_nth_combination(lineup_no)
player_names = get_player_list(lineup, team.teamno)
interval = calculate_interval(team, lineup_no)
lineup_data = get_lineup_data(teamname, player_names, interval, type="none")
if lineup_data.empty:
return pd.Series()
# print lineup_data
row_s = lineup_data # This transforms the columns name used by stats.nba.com to the convention
# I used throughout the project
# row_s.rename(index={
# 'PLUS_MINUS': 'PTD',
# 'PTS': 'P',
# 'FT_PCT': 'FTP',
# 'FG_PCT': 'FGP',
# 'OREB': 'ORB',
# 'DREB': 'DRB',
# 'REB': 'TRB',
# 'FG3_PCT': 'B3P',
# 'PF': 'F'
# }, inplace=True)
# row_s.drop("PFD", inplace=True) # drop Personal Foul Drawn since not used in our study
row_s.name = lineup_no
row_s["interval"] = interval
# if row_s.TYPE == ("median" or "mean"):
# pass
# else:
gamestats.append(row_s.T)
# print gamestats
game_agg = pd.DataFrame(gamestats).sum() # sum up a games all scores
# this PTD is still the actual PTD of the game, it is only the
# stats that have been replaced.
# game_agg.PTD = team.PTD
game_agg["Actual PTS"] = team.P
game_agg["Actual PTS scaled"] = team.P * game_agg.interval / float(team.interval)
game_agg["Actual PTD scaled"] = team.PTD * game_agg.interval / float(team.interval)
game_agg["Actual PTD"] = team.PTD
game_agg = game_agg.append(opponent_team_mean["OPP_FGM":])
# print game_agg
return game_agg
def opplineup(filename):
# namecolumns = ['a1', 'a2', 'a3', 'a4', 'a5', 'h1', 'h2', 'h3', 'h4', 'h5']
DATA_FOR_NOT_FOUND = "median"
out = []
df = pd.read_csv(PLAY_BY_PLAY_DIRECTORY + filename)
team_name = filename[9:12]
opp_team_name = filename[12:15]
if (opp_team_name not in ALLTEAMS) and (team_name not in ALLTEAMS):
print '%s, one of the team not available, skip!' % filename
return pd.DataFrame()
else:
add_aid_rows(df, [team_name, opp_team_name])
total_lineup = df.head(1).lineup_no.item() + 1
for lineup_no in range(1, int(total_lineup)): # go through all the lineups in the game
lineup_df = df[df.lineup_no == lineup_no] #
interval = lineup_df.head(1).timeleft.item() - lineup_df.tail(1).timeleft.item() # calculate the
# interval of this lineup
# print 'This is lineup: %s, the interval is %s' %(lineup_no,interval)
team1 = lineup_df[lineup_df.team == team_name]
team2 = lineup_df[lineup_df.team == opp_team_name]
for team_no, (team_name, team_data), in enumerate(zip([team_name, opp_team_name], [team1, team2])):
if team_data.empty: # check that the lineup contains this data of this team
print '%s, %s, %s, team not in this lineup data' % (filename, lineup_no, team_name)
else:
team_players = get_player_list(team_data, team_no + 1)
lineup_data = get_lineup_data(team_name, team_players, type="none") # retrieve data from stats.nba
if lineup_data.empty is False:
team_points = score_calculate(team_data)
lineup_data['interval'] = interval # calculate the interval of this lineup
lineup_data['PPM'] = float(team_points) / interval * 60 # calculate PPM(POINTS PER MINUTE)
# print lineup_data
out.append(lineup_data) # append a line to the lineup data
# print pd.concat(out, axis=1).T
if out != []:
return pd.concat(out, axis=1).T # concatenate all lineup data of one game
else:
return pd.Series()
