def test_get_stats(self):
df = get_stats('LaMarcus Aldridge')
expected_columns = ['Season', 'Age', 'Tm', 'Lg', 'Pos', 'G', 'GS', 'MP', 'FG', 'FGA', 'FG%', '3P', '3PA', '3P%', '2P', '2PA', '2P%', 'eFG%', 'FT', 'FTA', 'FT%', 'ORB', 'DRB', 'TRB', 'AST', 'STL', 'BLK', 'TOV', 'PF', 'PTS']
self.assertListEqual(list(df.columns), expected_columns)
df = get_stats('LaMarcus Aldridge', career=True)
expected_columns = ['Season', 'Tm', 'Lg', 'G', 'GS', 'MP', 'FG', 'FGA', 'FG%', '3P', '3PA', '3P%', '2P', '2PA', '2P%', 'eFG%', 'FT', 'FTA', 'FT%', 'ORB', 'DRB', 'TRB', 'AST', 'STL', 'BLK', 'TOV', 'PF', 'PTS']
self.assertListEqual(list(df.columns), expected_columns)
df = get_stats('LaMarcus Aldridge', playoffs=True, career=True)
expected_columns = ['Season', 'Tm', 'Lg', 'G', 'GS', 'MP', 'FG', 'FGA', 'FG%', '3P', '3PA', '3P%', '2P', '2PA', '2P%', 'eFG%', 'FT', 'FTA', 'FT%', 'ORB', 'DRB', 'TRB', 'AST', 'STL', 'BLK', 'TOV', 'PF', 'PTS']
self.assertListEqual(list(df.columns), expected_columns)
def test_get_stats(self):
expected_columns = [
'SEASON', 'AGE', 'TEAM', 'LEAGUE', 'POS', 'G', 'GS', 'MP', 'FG',
'FGA', 'FG%', '3P', '3PA', '3P%', '2P', '2PA', '2P%', 'eFG%', 'FT',
'FTA', 'FT%', 'ORB', 'DRB', 'TRB', 'AST', 'STL', 'BLK', 'TOV',
'PF', 'PTS'
]
df = get_stats('LaMarcus Aldridge')
self.assertCountEqual(list(df.columns), expected_columns)
df = get_stats('LaMarcus Aldridge', career=True)
self.assertCountEqual(list(df.columns), expected_columns)
df = get_stats('LaMarcus Aldridge', playoffs=True, career=True)
self.assertCountEqual(list(df.columns), expected_columns)
def player():
if request.method == 'GET':
return f"The URL /player is accessed directly. Try going to '/' to search for player"
form_data = request.form
name = form_data['player']
year = form_data['year']
name = name.title()
try:
stats = get_stats(name, 'PER_GAME', ask_matches=False)
except:
return render_template('error.html')
pl = stats[[
'SEASON', 'TEAM', 'PTS', 'TRB', 'AST', 'STL', 'BLK', 'FG%', '3P%'
]]
cur_season = pl.loc[pl['SEASON'] == year]
pic = get_player_headshot(name, ask_matches=False)
return render_template('base.html',
data=cur_season.to_html(
classes='table table-striped',
index=False,
col_space=100,
justify='match-parent'),
name=name,
pic=pic,
year=year)
def view_mvp_candidates():
mvp_candidates = [
'LeBron James', 'Giannis Antetokounmpo', 'James Harden',
'Anthony Davis', 'Kawhi Leonard', 'Jimmy Butler', 'Chris Paul'
]
display_rank = 1
for mvp_candidate in mvp_candidates:
print(f"Our Ranking #{display_rank}: {mvp_candidate}")
display_rank += 1
print()
mvp_candidate = input("Whose stats would like to see? ")
# ADD Check Condition Here.
# Loop if Player not one of the above.
# Error when entry is 1.
game_logs = get_game_logs(mvp_candidate,
'2019-10-22',
'2020-03-01',
playoffs=False)
game_stats = get_stats(mvp_candidate,
stat_type='PER_GAME',
playoffs=False,
career=False)
print(f"-----------{mvp_candidate}'s 2019-2020 GAME LOGS-----------")
print(game_logs)
print(f"-----------{mvp_candidate}'s CAREER GAME STATS-----------")
print(game_stats)
def scrap_roster(roster=None, suffixes=None):
if roster is not None:
for raw_player_name in roster["PLAYER"]:
try:
player_name = clean_player_name(raw_player_name)
player_data = get_stats(player_name, suffixes=suffixes)
print(player_data)
except:
scrap_failures.append(raw_player_name)
def player_stats(player, stat_type, playoffs, career):
print("Saving into excel sheet... ")
writer = ExcelWriter("./Excel-Sheets/%s_STATS__%s_REGULAR.xlsx" % (player.upper(), stat_type))
sheet_name = player.upper() + "- REGULAR"
if playoffs:
writer = ExcelWriter("./Excel-Sheets/%s_STATS_%s_PLAYOFFS.xlsx" % (player.upper(), stat_type))
sheet_name = player.upper() + "-PLAYOFFS"
player_stats_df = get_stats(player, stat_type, playoffs, career)
player_stats_df.to_excel(writer, sheet_name=sheet_name, startrow=0, startcol=0, index=False, header=True)
writer.save()
def get_missing_win_shares(injury_report, team):
team_injuries = injury_report.loc[injury_report['TEAM'] == team]
team_injuries = team_injuries.loc[team_injuries['STATUS'] == 'Out']
team_injuries = list(team_injuries['PLAYER'])
win_shares = 0
for player in team_injuries:
try:
player_stats = get_stats(player, stat_type='ADVANCED').tail(1)
if not player_stats.loc[player_stats['SEASON'] == '2019-20'].empty:
win_shares += list(player_stats['WS'])[0]
except:
pass
return round(win_shares)
def get_player_df(players, stat_type, stat, playoffs, player_labels):
"""
Get dataframe from basketball_scraper
:param players: list of player names
:return:
"""
if not player_labels:
player_labels = players
players_df = []
for player, player_label in zip(players, player_labels):
player_df = bball_players.get_stats(player, stat_type, playoffs)
player_df = df_format(player_df, stat)
players_df.append((player_df, player_label))
return players_df
def get_playoff_minutes(row):
p = row['PLAYER']
print('processing: ' + p)
# fix encoding issue from bballref table scrape
# č = ć, but is coming in as just Ä, which is invalid
try:
player = p.encode('cp1252').decode()
except:
print('encoding error ' + p)
index = p.find('Ä')
while index > -1 and index <= len(p):
try:
s = 'Ä' + p[index + 1]
s.encode('cp1252').decode()
index += 1
except:
p = p[:index + 1] + '‡' + p[index + 1:]
index = p.find('Ä', index)
player = p.encode('cp1252', 'ignore').decode()
print('encoding error fixed: ' + player)
try:
player_df = get_stats(player,
stat_type='TOTALS',
playoffs=True,
career=True,
ask_matches=False)
if not player_df.empty:
return player_df['MP'].sum()
except (ValueError, IndexError):
return 0
except:
print('api error ' + player)
return 0
def user_input():
team_or_player = input("What stats would you like to see: (1)Team or (2)Player ")
if team_or_player == '1':
user_team = input("What team would you like to view? ")
user_season = int(input("Enter the desired season: "))
user_stats = input("Enter the number for what stat you would like to see: \n1.) Roster List\n"
"2.) Team Stats\n"
"3.) Team Opponent Stats\n"
"4.) Roster Stats\n"
"5.) Misc Stats\n")
print_or_excel = input("Would you like to save information in an excel sheet (Y/N)? ").upper()
if user_stats == '1':
if print_or_excel == 'Y':
roster_list(user_team, user_season)
print("Saved")
elif print_or_excel == 'N':
print(get_roster(user_team, user_season))
elif user_stats == '2':
data_format = input("Enter data format: (TOTAL | PER_GAME): ").upper()
if print_or_excel == 'Y':
team_stats(user_team, user_season, data_format)
print(get_team_stats(user_team, user_season, data_format))
elif print_or_excel == 'N':
print(get_team_stats(user_team, user_season, data_format))
elif user_stats == '3':
data_format = input("Enter data format: (TOTAL | PER_GAME: ").upper()
if print_or_excel == 'Y':
team_opp_stats(user_team, user_season, data_format)
print(get_opp_stats(user_team, user_season, data_format))
elif print_or_excel == 'N' or 'n':
print(get_opp_stats(user_team, user_season, data_format))
elif user_stats == '4':
user_playoffs = input("Would you like playoff roster stats? (Y/N) ")
data_format = input("Enter data format: (TOTAL | PER_GAME): ").upper()
playoffs = False
if user_playoffs == 'Y':
playoffs = True
if print_or_excel == 'Y':
roster_stats(user_team, user_season, data_format, playoffs)
print(get_roster_stats(user_team, user_season, data_format, playoffs))
elif print_or_excel == 'N':
print(get_roster_stats(user_team, user_season, data_format, playoffs))
elif user_stats == '5':
if print_or_excel == 'Y':
team_misc(user_team, user_season)
print(get_team_misc(user_team, user_season))
elif print_or_excel == 'N':
print(get_team_misc(user_team, user_season))
elif team_or_player == '2':
user_player = input('What player do you like to view? ')
print_or_excel = input("Save into excel sheet? (Y/N): ").upper()
user_stats = input("Enter the number for what stat you would like to see: \n"
"1.) Player Stats\n"
"2.) Game Logs\n")
user_playoffs = input("Playoff Stats? (Y/N): ").upper()
if user_stats == '1':
stat_type = input("Enter stat type: (PER_GAME | PER_MINUTE | PER_POSS | ADVANCED): ").upper()
user_career = input("Career Stats? (Y/N): ").upper()
if user_playoffs == 'Y':
playoffs = True
else:
playoffs = False
if user_career == 'Y':
career = True
else:
career = False
if print_or_excel == "Y":
player_stats(user_player, stat_type, playoffs, career)
print(get_stats(user_player, stat_type, playoffs, career))
elif print_or_excel == "N":
print(get_stats(user_player, stat_type, playoffs, career))
elif user_stats == '2':
user_start_date = input("Enter start date: (YYYY-MM-DD): ").upper()
user_end_date = input("Enter end date: (YYYY-MM-DD): ")
if user_playoffs == 'Y':
playoffs = True
else:
playoffs = False
if print_or_excel == "Y":
player_game_logs(user_player, user_start_date, user_end_date, playoffs)
print(get_game_logs(user_player, user_start_date, user_end_date, playoffs))
elif print_or_excel == "N":
print(get_game_logs(user_player, user_start_date, user_end_date, playoffs))
#Get team roster stats
bulls_2020 = get_roster_stats('CHI', 2020)
#Select individual player
bulls_2020[bulls_2020.PLAYER == "Coby White"]
''' Get individual players
name - Player full name (e.g. 'LaMarcus Aldridge')
stat_type - One of 'PER_GAME', 'PER_MINUTE', 'PER_POSS', 'ADVANCED'
playoffs - Whether to return Playoff stats or not. One of True|False. Default value is False
career - Whether to return career stats or not. One of True|False. Default value is False
'''
coby = get_stats("Coby White",
stat_type='ADVANCED',
playoffs=False,
career=False)
coby.head(10)
d = get_player_headshot('Coby White')
d = get_shot_chart('2019-12-28', 'TOR', 'BOS')
#Import college players
from sportsreference.ncaab.roster import Player as pl
player_ncaa = pl('kenyon-martin-1')
#get available object variables
dir(player_ncaa)
player_ncaa.dataframe.iloc[-2, :]
#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_team_stats('GSW', 2019, data_format='PER_GAME')
print(s)
s = get_opp_stats('GSW', 2019, data_format='PER_GAME')
print(s)
s = get_roster_stats('GSW', 2019, data_format='PER_GAME', playoffs=False)
print(s)
s = get_team_misc('GSW', 2019)
print(s)
from basketball_reference_scraper.players import get_stats, get_game_logs
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
#Documentation for library
#https://github.com/vishaalagartha/basketball_reference_scraper
#dependencies
import pandas as pd
import numpy as np
import os
from basketball_reference_scraper.players import get_stats, get_game_logs, get_player_headshot
import scipy.stats as st
import matplotlib.pyplot as plt
# GEt stats
player_input = input("Please Enter an NBA Player: ")
player_stats = get_stats(player_input)
#Graphing
x = (player_stats["MP"])
xx = x.dropna()
y = player_stats["FGA"]
yy = y.dropna()
plt.scatter(x, y)
plt.xlabel("Minutes")
plt.ylabel("Points")
#Linear Regression
#import scipy.stats as st
cor = round(st.pearsonr(x, y)[0], 2)
print(f"The correlation between minutes and game score is {cor}")
model = st.linregress(xx, yy)
