def scrape_and_clean_game(date, away, home, year):
# scrape game
df = get_pbp(date, away, home)
# playoffs?
if (int(date[5:7]) >=
5) and (int(date[8:10]) >=
22): # playoff start date; TODO: change to auto detect
playoffs = "Playoffs"
else:
playoffs = "Regular"
# change colnames and add new columns
df.columns = [
'quarter', 'time_remaining', 'away_action', 'home_action',
'away_score', 'home_score'
]
df.insert(0, 'home_team', [home for i in range(len(df.index))])
df.insert(0, 'away_team', [away for i in range(len(df.index))])
df.insert(0, 'playoffs', [playoffs for i in range(len(df.index))])
df.insert(0, 'month', [date[5:7] for i in range(len(df.index))])
season = str(int(year) - 1) + "-" + year
df.insert(0, 'season', [season for i in range(len(df.index))])
df.insert(0, 'date', [date for i in range(len(df.index))])
gameId = (date + away + home).replace('-', '')
df.insert(0, 'gameId', [gameId for i in range(len(df.index))])
return df
def scrape_and_clean_game(date, away, home, playoffs, year):
# scrape game
df = get_pbp(date, away, home)
# change colnames and add new columns
df.columns = [
'quarter', 'time_remaining', 'away_action', 'home_action',
'away_score', 'home_score'
]
df.insert(0, 'home_team', [home for i in range(len(df.index))])
df.insert(0, 'away_team', [away for i in range(len(df.index))])
df.insert(0, 'playoffs', [playoffs for i in range(len(df.index))])
df.insert(0, 'month', [date[5:7] for i in range(len(df.index))])
season = str(int(year) - 1) + "-" + year
df.insert(0, 'season', [season for i in range(len(df.index))])
df.insert(0, 'date', [date for i in range(len(df.index))])
gameId = (date + away + home).replace('-', '')
df.insert(0, 'gameId', [gameId for i in range(len(df.index))])
return df
#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""
Created on Thu Jan 30 14:43:11 2020
@author: Nillan
"""
from basketball_reference_scraper.shot_charts import get_shot_chart
from basketball_reference_scraper.pbp import get_pbp
d = get_shot_chart('2019-12-28', 'TOR', 'BOS')
pbp = get_pbp('2019-12-28', 'TOR', 'BOS')
# when joel embiid is making more threes, how does bill simmons do
#by qtr?
def test_pbp(self):
df = get_pbp('2020-01-06', 'DEN', 'ATL')
expected_columns = ['QUARTER', 'TIME_REMAINING', 'DENVER_ACTION', 'ATLANTA_ACTION', 'DENVER_SCORE', 'ATLANTA_SCORE']
self.assertListEqual(list(df.columns), expected_columns)
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)
from basketball_reference_scraper.shot_charts import get_shot_chart
s = get_shot_chart('2020-01-13', 'CHI', 'BOS')
print(s)
from basketball_reference_scraper.injury_report import get_injury_report
s = get_injury_report()
print(s)
def test_pbp(self):
df = get_pbp('2020-01-06', 'DEN', 'ATL')
expected_columns = ['Clock', 'Denver', 'Atlanta', 'Denver Score', 'Atlanta Score']
self.assertListEqual(list(df.columns), expected_columns)
def main():
"""Generate a summary email for last night's NBA games"""
# loop through last night's games
yest_month = (datetime.now() - timedelta(1)).strftime('%B').lower()
schedule = get_season_schedule(yest_month)
# save df to create each plot
plot_dfs = []
plot_titles = []
for i, row in schedule.iterrows():
if row.Date2 == (datetime.now() - timedelta(1)).strftime('%Y-%m-%d'):
date = row.Date2
away = row.Visitor
home = row.Home
pbp = get_pbp(date, away, home).fillna('no action')
dfs_box = get_box(date, away,
home) # returns list of [away_df, home_df]
df_away_raw = dfs_box[0]
df_home_raw = dfs_box[1]
# cut 'out' players and change columns to numeric
trimmed_dfs = cols_to_numeric(
cut_players_out(df_away_raw, df_home_raw, away, home))
df_away = trimmed_dfs[0]
df_home = trimmed_dfs[1]
# print team scores and top performers
print_team_scoring_and_top_performers(df_away, df_home, away, home)
# free throws
print_free_throws(df_away, df_home, away, home)
# 3 pointers
print_three_pointers(df_away, df_home, away, home)
# shots at rim
shots_at_rim = count_shots_at_rim(pbp, away, home)
write_to_txt([shots_at_rim[8], shots_at_rim[9]])
# rebounds
print_rebounding(df_away, df_home, away, home)
# possessions and OT
print_possessions_and_ot(pbp)
# add link to highlights
url = "https://www.youtube.com/playlist?list=PLlVlyGVtvuVkIjURb-twQc1lsE4rT1wfJ"
write_to_txt([f"Game highlights here: {url} \n"])
# save plot df for current game
plot_dfs.append(plot_score_vs_time(pbp))
plot_titles.append(f"{away} v {home}")
print("Sleeping for 3 seconds... \n")
print("\n")
time.sleep(3)
# space between games in the text file
write_to_txt(["\n"])
"""
# plot score vs. time for each game
# fig, axs = plt.subplots(x,y) [2,2] for 4, [3,2] for 6, [3,3] for 9 etc.
num_games = len(plot_dfs)
print(f"num_games: {num_games}")
x, y = 0, 0
x_len = int((num_games - 0.01)**0.5) + 1
y_len = int((num_games - x_len) / (x_len + 0.01)) + 2
print(f"x_len: {x_len}")
print(f"y_len: {y_len}")
fig, axs = plt.subplots(x_len, y_len)
for df, title in zip(plot_dfs, plot_titles):
x1 = df.iloc[:,0]
x2 = df.iloc[:,1]
_y = df.index
axs[x,y].plot(x1, _y)
axs[x,y].plot(x2, _y)
axs[x,y].set_ylabel("Points")
axs[x,y].set_title(title)
axs[x,y].set_xlabel("Minutes played")
axs[x,y].set_xticks([0, 12, 24, 36, 48])
axs[x,y].grid('on')
if x == x_len - 1:
x = 0
y += 1
else:
x += 1
fig.tight_layout()
plt.savefig(f'game_flow_charts/gameflows_{date}.png')
print("Game flows saved")
"""
print("Finished")