def team_array():
'''Returns a dictionary of pd dataframe objects. Filters out the data
to only include a specific team. Ex usage:
df = team_array()
print df['Dallas Cowboys']
'''
table = load.getDataset()
teams = get_teams(table)
team_keys = get_team_keys()
combined = {}
for team in team_keys:
combined[team] = pd.merge(teams[team + ' home'], \
teams[team + ' away'], how = 'outer')
return combined
def write_teams():
table = load.getDataset()
teams_table = separate_teams(table)
team_keys = teams_table.keys()
team_keys = list(set([x[:-5] for x in team_keys]))
yes_ser = pd.Series(np.repeat(np.array(['True']), len(table)))
no_ser = pd.Series(np.repeat(np.array(['False']), len(table)))
for team in team_keys:
home_team = teams_table[team + ' home']
away_team = teams_table[team + ' away']
home_team = format_home(home_team)
home_team['home_field?'] = yes_ser
away_team = format_away(away_team)
away_team['home_field?'] = no_ser
table_out = pd.merge(home_team, away_team, how = 'outer')
dirpath = os.getcwd()[:-10] + '/data/teamdata/'
table_out.to_csv(dirpath + team + '.csv')
def validate():
'''basic sanity checks on the numerical parts of the data'''
table = load.getDataset()
# These are inequalities
for team in ['ht_', 'at_']:
for elem in [('rush_attempts', 'rush_TDs'), ('pass_attempt', 'pass_comp'),
('pass_comp', 'pass_TDs'), ('fumbles', 'fumbles_lost'),
('3rd_down_attempts', '3rd_down_converted'),
('4th_down_attempts', '4th_down_converted')]:
test = table[team + elem[0]] >= table[team + elem[1]]
assert(len(set(test)) == 1)
# These are equalities
for elem in [('net_pass_yards', 'rush_yards', 'total_yards'),
('fumbles_lost', 'INT', 'turnovers')]:
test = table[team + elem[0]] + table[team + elem[1]] - \
table[team + elem[2]]
assert(len(set(test)) == 1)
home_TOP = np.array([int(x.split(':')[0])*60 + \
int(x.split(':')[1]) for x in table['ht_TOP']])
away_TOP = np.array([int(x.split(':')[0])*60 + \
int(x.split(':')[1]) for x in table['at_TOP']])
assert(len(set(home_TOP + away_TOP >= 3595)) == 1 )
def home_away_differences():
''' Finds the differences between teams and their opponents based on
location. Returns a dictionary of numpy arrays. Ex. use:
a = home_away_differences()
plt.hist(a['Dallas Cowboys home score'], 20)
which would plot a hist of how many more points the Cowboys have vs their
opponent when the Cowboys are playing at home.
a['Dallas Cowboys away score'] would be their stats when they are away
Might delete. Too specific of a function
'''
table =load.getDataset()
all_teams = get_teams(table)
team_keys = get_team_keys()
columns = list(table.keys())
# ghetto, hardcoded filtering method :(
columns = [x[3:] for x in columns if ('at' in x)]
columns.remove('endance')
columns.remove('TOP')
stat_diff = {}
for team in team_keys:
for stat in columns:
for location in [(' home', 'ht_', 'at_'), (' away', 'at_', 'ht_')]:
stat_diff[team + location[0] + ' ' + stat] = \
np.array(all_teams[team + location[0]][location[1] + stat]) - \
np.array(all_teams[team + location[0]][location[2] + stat])
return stat_diff
def cum_array(stat):
'''Returns a pd series of a particular stat from the table. Input is
a string'''
return load.getDataset()[stat]
def cum_stat(stat):
'''Gives a rough idea of the dsistribution of a stat for all games at
a glance. Input is a string'''
table = load.getDataset()
return table.describe()[stat]
def get_team_keys():
'''gets a list of all teams'''
table = load.getDataset()
teams = list(set(list(table['home_team'].values) + list(table['away_team'].values)))
teams = [x.strip() for x in teams]
return teams
