def test_get_standings(self):
d = get_standings()
self.assertListEqual(list(d.keys()), ['EASTERN_CONF', 'WESTERN_CONF'])
df = d['WESTERN_CONF']
expected_columns = ['TEAM', 'W', 'L', 'W/L%', 'GB', 'PW', 'PL', 'PS/G', 'PA/G']
self.assertListEqual(list(df.columns), expected_columns)
def get_team_standings(self, subset_by_seasons: list = None):
"""Assumptions : the season is over by June 1st.
TODO : Use the season dataset to find last game date.
"""
year = datetime.datetime.now().year
allowed_seasons = range(1974, year + 1)
if subset_by_seasons is not None:
seasons = [
season for season in subset_by_seasons
if season in allowed_seasons
]
else:
seasons = allowed_seasons
total_dfs = []
for season in seasons:
print("Retrieving standings of season", season, "...")
date = "06-01-" + str(season)
dfs = []
results = get_standings(date=date)
for conference, data in results.items():
data = data.dropna(axis="index", how="any")
data = data.sort_values(by="W/L%", ascending=False)
data = data.reset_index(drop=True)
data.loc[:, "CONF"] = conference
data.loc[:, "CONF_RANK"] = data.index + 1
data.loc[:, "TEAM"] = data["TEAM"].str.upper().str.replace(
"[^A-Z]", "")
team_names = {}
for raw, short in self.team_names.items():
raw = "".join(filter(str.isalpha, raw)).upper()
team_names[raw] = short
data = data[~data["TEAM"].str.contains("DIVISION")]
unmapped_teams = [
team for team in data["TEAM"].unique()
if team not in team_names.keys()
]
data.loc[:, "TEAM"] = data["TEAM"].map(team_names)
if data["TEAM"].isna().sum() > 0:
raise ValueError("Unknown/unmapped teams : %s",
unmapped_teams)
data.loc[:, "GB"] = (data["GB"].str.replace("—", "0.0").astype(
float, errors="raise"))
data.loc[:, "TEAM_SEASON"] = data["TEAM"] + "_" + str(season)
data.loc[:, "SEASON"] = season
data = data.set_index("TEAM_SEASON", drop=True)
dfs.append(data)
all_conf_df = pandas.concat(dfs,
join="outer",
axis="index",
ignore_index=False)
total_dfs.append(all_conf_df)
all_conf_df = pandas.concat(total_dfs,
join="outer",
axis="index",
ignore_index=False)
return all_conf_df
def test_get_standings(self):
d = get_standings()
self.assertListEqual(list(d.keys()),
['Eastern Conference', 'Western Conference'])
df = d['Western Conference']
expected_columns = [
'Western Conference', 'W', 'L', 'W/L%', 'GB', 'PW', 'PL', 'PS/G',
'PA/G'
]
self.assertListEqual(list(df.columns), expected_columns)
def get_team_list():
from basketball_reference_scraper.seasons import get_standings
df = get_standings("2019-01-01")
east = df['EASTERN_CONF']
west = df['WESTERN_CONF']
east2 = east.TEAM.str.rstrip("*").to_list()
west2 = west.TEAM.str.rstrip("*").to_list()
for i in range(0, len(west2)):
east2.append(west2[i])
return east2
#Script to get NBA roster from a certain year
from basketball_reference_scraper.teams import get_roster
roster = get_roster('CLE', 2016)
print("\nCleveland Cavaliers 2016 Roster:")
print(roster)
#Script to retrieve an indiviual player's stats by year
from basketball_reference_scraper.players import get_stats
stats = get_stats('Collin Sexton', stat_type='PER_GAME')
print("\nCollin Sexton Stats:")
print(stats)
#Script to get NBA standing on a certain date
from basketball_reference_scraper.seasons import get_standings
standings = get_standings()
print("\nCurrent NBA Standings:")
print(standings)
s = get_stats('Stephen Curry',
stat_type='PER_GAME',
playoffs=False,
career=False)
print(s)
df = get_game_logs('LeBron James', '2010-01-19', '2014-01-20', playoffs=False)
print(df)
from basketball_reference_scraper.seasons import get_schedule, get_standings
s = get_schedule(2018, playoffs=False)
print(s)
s = get_standings(date='2020-01-06')
print(s)
from basketball_reference_scraper.box_scores import get_box_scores
s = get_box_scores('2020-01-13',
'CHI',
'BOS',
period='GAME',
stat_type='BASIC')
print(s)
from basketball_reference_scraper.pbp import get_pbp
s = get_pbp('2020-01-13', 'CHI', 'BOS')
print(s)
start_year = 2010
end_year = 2013
df_columns = ['year', 'team', 'b2b_away_pct', 'b2b_away_wins', 'b2b_away_games', 'season_WL_pct', 'diff']
final_dict = {}
flat_final = pd.DataFrame(columns = df_columns)
#year = 2010
for year in range(start_year, end_year, 1):
all_results = get_schedule(year, playoffs=False)
teams = all_results.HOME.unique().tolist()
date = str(year) + '-06-06'
standings = get_standings(date=date)
combined_standings = pd.concat([standings['EASTERN_CONF'], standings['WESTERN_CONF']])
combined_standings.TEAM = combined_standings.TEAM.str.rstrip('*')
final_dict[year] = {}
for team in teams:
#team = 'Miami Heat'
heat_home = all_results[all_results.HOME.eq(team)]
heat_away = all_results[all_results.VISITOR.eq(team)]
heat = all_results[(all_results['HOME']==team) | (all_results['VISITOR']==team)]
num_games = heat.shape[0]
