def scrape_nba_results():
''' Scrape recent NBA results'''
url = 'http://www.betexplorer.com/basketball/usa/nba/results/'
df = pd.read_html(get(url).text)[0]
homeTeam = df[0].apply(lambda r: str.split(r, sep='-')[0].strip())
homeTeam[homeTeam == 'Portland Trail Blazers'] = 'Portland Trailblazers'
homePoints = df[1].apply(lambda r: str.split(r, sep=':')[0].strip())
awayTeam = df[0].apply(lambda r: str.split(r, sep='-')[1].strip())
awayTeam[awayTeam == 'Portland Trail Blazers'] = 'Portland Trailblazers'
awayPoints = df[1].apply(lambda r: str.split(r, sep=':')[1].strip())
awayPoints = awayPoints.apply(lambda r: str.split(r, sep='ET')[0].strip())
dates = df[4].apply(lambda r: datetime.strptime(r, '%d.%m.%Y'))
# The dates on this website are GMT so are one day advanced.
dates = dates.apply(lambda r: datetime.strftime(r - timedelta(days=1),
'%d/%m/%Y'))
df['Date'] = dates
df['HomeTeam'] = homeTeam
df['AwayTeam'] = awayTeam
df['HomePoints'] = homePoints
df['AwayPoints'] = awayPoints
df['HomeWin'] = homePoints > awayPoints
df = df.ix[:, 5:11]
teams = lookup_teams()
df['HomeId'] = df.merge(df.merge(teams, left_on='HomeTeam',
right_on='Franchise', sort=False))['TeamId']
df['AwayId']= df.merge(df.merge(teams, left_on='AwayTeam',
right_on='Franchise', sort=False))['TeamId']
return df
def scrape_model_probs():
'''Scrape CARM-Elo probabilities for next round of matches'''
url = 'http://projects.fivethirtyeight.com/2016-nba-picks/'
df = pd.read_html(get(url).text)[1].ix[:, 1:5]
df = df[df.ix[:, 0] != 'X']
df.columns = ['AwayTeam', 'AwayModelProb', 'HomeModelProb', 'HomeTeam']
def parse_probability(string):
return float(str.split(string, sep='%')[0])/100
df['AwayModelProb'] = df['AwayModelProb'].apply(parse_probability)
df['HomeModelProb'] = df['HomeModelProb'].apply(parse_probability)
teams = lookup_teams()
away = pd.merge(df, teams, left_on='AwayTeam', right_on='Abbreviation538')
home = pd.merge(df, teams, left_on='HomeTeam', right_on='Abbreviation538')
df['Date'] = datetime.today().strftime('%d/%m/%Y')
df['AwayId'] = away['TeamId']
df['AwayTeam'] = away['Franchise']
df['HomeId'] = home['TeamId']
df['HomeTeam'] = home['Franchise']
df = df[['Date', 'HomeId', 'AwayId', 'HomeTeam', 'AwayTeam',
'HomeModelProb', 'AwayModelProb']]
return df
