#hockey-reference.com associates team names with an abbreviation for individual team data links. The abbreviations they used were not standarized or could be found on website so I manually made a dictionary that associates the team
teams_with_abr = {'St. Louis Blues':'STL', 'Detroit Red Wings':'DET', 'Philadelphia Flyers' : 'PHI',
'New Jersey Devils' : 'NJD','Washington Capitals':'WSH', 'Dallas Stars':'DAL',
'Toronto Maple Leafs':'TOR', 'Florida Panthers' :'FLA', 'Colorado Avalanche' : 'COL',
'Ottawa Senators' : 'OTT' , 'Los Angeles Kings' : 'LAK' , 'Phoenix Coyotes' : 'PHX', 'Arizona Coyotes' : 'ARI',
'Pittsburgh Penguins' : 'PIT', 'Edmonton Oilers' : 'EDM' , 'San Jose Sharks' : 'SJS',
'Buffalo Sabres' : 'BUF', 'Carolina Hurricanes' : 'CAR', 'Vancouver Canucks' : 'VAN',
'Montreal Canadiens' : 'MTL' , 'Mighty Ducks of Anaheim' : 'MDA', 'Chicago Blackhawks' : 'CHI',
'Calgary Flames' : 'CGY' , 'New York Rangers' : 'NYR', 'Boston Bruins' :'BOS', 'Nashville Predators' : "NSH",
'New York Islanders' : 'NYI' , 'Tampa Bay Lightning' : 'TBL' , 'Atlanta Thrashers' : 'ATL' , 'Winnipeg Jets' : 'WPG',
'Vegas Golden Knights' : 'VEG', 'Seattle Kraken' : 'SEA', 'Anaheim Ducks' : 'ANA', 'Minnesota Wild' : 'MIN'
,'Columbus Blue Jackets':'CBJ'}
#all my data is from the year 2000 to 2022. However, the current 2022 season has not been
#finished so this data will be saved separately to later make predictions on who will win the stanley cup this year.
years = [i for i in range(2000,2023)]
regular_season = []
#saving the dataframe for current 2022 regular season separately
current_regular_season =[]
#looping through each year
for year in years:
# the nhl was not held in 2005 so ignore this year
if (year != 2005 ):
#had to donwnload data manually and then read it from a dictionary
data = f"data/nhl_{year}.csv"
#reading in csv file as pandas dataframe
curr_df = pd.read_csv(data,index_col=[0])
#had to tidy up some of the columns to make it more readable
curr_df = curr_df.rename(columns = {'Unnamed: 1':'blah'})
curr_df.columns = curr_df.iloc[0]
curr_df = curr_df.drop(curr_df.index[0])
curr_df = curr_df
curr_df = curr_df.rename(columns = {np.NaN:'Team'})
curr_df.columns
curr_df = curr_df[:-1]
# a * in a teams name means that are a playoff Team. So only want to save teams data from regular season that are playoff teams
#right before submitting project, the website updated and removed the * for each
#team that made the 2022 playoffs so had to add in this check to get the 16 teams with highest points
if (year == 2022):
curr_df = curr_df.iloc[0:16]
else:
curr_df = curr_df.loc[curr_df['Team'].str.contains("\*")]
#after filtering can strip the * from each teams name since only looking at playoff teams
curr_df['Team'] = curr_df['Team'].str.strip("*")
#converting desired columns into floats (previously were saved as strings)
curr_df[['AvAge', 'PTS%', 'PIM/G', 'S%', 'SV%','SRS', 'SOS', 'GF/G', 'GA/G', 'PP%', 'PK%']] = curr_df[['AvAge', 'PTS%', 'PIM/G', 'S%', 'SV%','SRS', 'SOS', 'GF/G', 'GA/G', 'PP%', 'PK%']].astype(float)
#converting desired columns into ints (previously were saved as strings)
curr_df[['W', 'L','OL','PTS','GF','GA']] = curr_df[['W', 'L','OL','PTS','GF','GA']].astype(int)
#creating additional columns that will be used later in project
curr_df['PlayoffWins'] = np.NAN
curr_df['WinPctLast10'] = np.NAN
curr_df['ID'] = np.NAN
curr_df['PlayoffRound'] = np.NAN
#getting a teams id, used because each teams website is based on their id
curr_df['ID'] = curr_df['Team'].map(teams_with_abr)
#getting each teams win_percentage in last 10 games (will be seen why later)
for tm in curr_df['ID']:
curr_team = pd.read_html(f"https://www.hockey-reference.com/teams/{tm}/{year}_games.html")
curr_team = curr_team[0].tail(10)
#for some reason the 2022 season column names are labeled differently
if (year == 2022):
curr_team = curr_team.rename(columns = {'Unnamed: 7':'Result'})
else:
curr_team = curr_team.rename(columns = {'Unnamed: 6':'Result'})
#gettings teams win losses and ties for last 10 games.
game_result = curr_team['Result'].value_counts()
#From 2000-2004 the NHL allowed ties so I had to check for each possible combination of this.
if 'T' in game_result.index and 'L' in game_result.index and 'W' in game_result.index:
win_percentage = game_result['W'] / (game_result['W'] + game_result['L'] + game_result['T'])
elif 'W' in game_result.index and 'L' in game_result.index:
win_percentage = game_result['W'] / (game_result['W'] + game_result['L'])
elif 'W' in game_result.index and 'T' in game_result.index:
win_percentage = game_result['W'] / (game_result['W'] + game_result['T'])
elif 'W' in game_result.index:
win_percentage = 1.0
else:
win_percentage = 0.0
#saving computed regular season win percentage in last 10 games to teams respective year
curr_df.loc[curr_df.loc[curr_df['ID'] == tm].index[0],['WinPctLast10']] = win_percentage
#saving current year
curr_df['year'] = year
curr_df = curr_df[['Team', 'AvAge', 'GP', 'W', 'L', 'OL', 'PTS', 'PTS%', 'GF',
'GF/G', 'GA/G', 'PP', 'PPO', 'PP%', 'PPA', 'PPOA','GA','S%', 'SRS', 'SOS',
'PK%', 'PIM/G','SV%','PlayoffWins', 'WinPctLast10', 'ID', 'year']]
#saved all the dataframes into an array of dataframes unless it is 2022 season since not yet completed
if (year != 2022):
regular_season.append(curr_df)
else:
current_regular_season.append(curr_df)
#example dataframe from the 2004 regular season
regular_season[4].head(100)
