passes_between['width'] = passes_between.pass_count / passes_between.pass_count.max() * max_line_width
average_locs_and_count['marker_size'] = (average_locs_and_count['count']
/ average_locs_and_count['count'].max() * max_marker_size)
##############################################################################
# Set color to make the lines more transparent when fewer passes are made
min_transparency = 0.3
color = np.array(to_rgba('white'))
color = np.tile(color, (len(passes_between), 1))
c_transparency = passes_between.pass_count / passes_between.pass_count.max()
c_transparency = (c_transparency * (1 - min_transparency)) + min_transparency
color[:, 3] = c_transparency
##############################################################################
# Plotting
pitch = Pitch(pitch_type='statsbomb', orientation='horizontal',
pitch_color='#22312b', line_color='#c7d5cc', figsize=(16, 11),
constrained_layout=True, tight_layout=False)
fig, ax = pitch.draw()
pass_lines = pitch.lines(passes_between.x, passes_between.y,
passes_between.x_end, passes_between.y_end, lw=passes_between.width,
color=color, zorder=1, ax=ax)
pass_nodes = pitch.scatter(average_locs_and_count.x, average_locs_and_count.y, s=average_locs_and_count.marker_size,
color='red', edgecolors='black', linewidth=1, alpha=1, ax=ax)
for index, row in average_locs_and_count.iterrows():
pitch.annotate(row.name, xy=(row.x, row.y), c='white', va='center', ha='center', size=16, weight='bold', ax=ax)
title = ax.set_title("{} {} Formation vs {}".format(team, formation, opponent), size=28, y=0.97, color='#c7d5cc')
fig.set_facecolor("#22312b")
df_shot_event.y,
df_shot_event.shot_end_x,
df_shot_event.shot_end_y,
comet=True,
label='shot',
color='#cb5a4c',
ax=ax)
# plot the angle to the goal
pitch.goal_angle(df_shot_event.x,
df_shot_event.y,
ax=ax,
alpha=0.2,
zorder=1.1,
color='#cb5a4c',
goal='right')
# plot the jersey numbers
for i, label in enumerate(df_freeze_frame.player_jersey_number):
pitch.annotate(label, (df_freeze_frame.x[i], df_freeze_frame.y[i]),
va='center',
ha='center',
color='white',
fontsize=12,
ax=ax)
# add a legend and title
legend = ax.legend(loc='center left', labelspacing=1, fontsize=15)
title = ax.set_title(
f'{df_shot_event.player_name.iloc[0]}\n{team1} vs. {team2}', fontsize=30)
stripe=True,
constrained_layout=False)
fig, ax = pitch.draw()
# plotting
ax.set_title('The first Game Messi played in the false 9 role',
fontsize=30,
pad=20)
# plot the kernel density estimation
pitch.kdeplot(df_false9.x, df_false9.y, ax=ax, cmap='plasma', linewidths=3)
# annotate
pitch.annotate('6-2 thrashing \nof Real Madrid', (25, 10),
color='white',
fontsize=25,
ha='center',
va='center',
ax=ax)
pitch.annotate('more events', (70, 30), (20, 30),
ax=ax,
color='white',
ha='center',
va='center',
fontsize=20,
arrowprops=dict(facecolor='white', edgecolor='None'))
pitch.annotate('fewer events', (51, 20), (20, 20),
ax=ax,
color='white',
ha='center',
va='center',
fontsize=20,
def createPassNetworks(match_data,
events_df,
matchId,
team,
max_line_width,
marker_size,
edgewidth,
dh_arrow_width,
marker_color,
marker_edge_color,
shrink,
ax,
kit_no_size=20):
# getting team id and venue
if match_data['home']['name'] == team:
teamId = match_data['home']['teamId']
venue = 'home'
else:
teamId = match_data['away']['teamId']
venue = 'away'
# getting opponent
if venue == 'home':
opponent = match_data['away']['name']
else:
opponent = match_data['home']['name']
# getting player dictionary
team_players_dict = {}
for player in match_data[venue]['players']:
team_players_dict[player['playerId']] = player['name']
# getting minute of first substitution
for i in events_df.index:
if events_df.loc[i, 'type'] == 'SubstitutionOn' and events_df.loc[
i, 'teamId'] == teamId:
sub_minute = str(events_df.loc[i, 'minute'])
break
# getting players dataframe
match_players_df = pd.DataFrame()
player_names = []
player_ids = []
player_pos = []
player_kit_number = []
for player in match_data[venue]['players']:
player_names.append(player['name'])
player_ids.append(player['playerId'])
player_pos.append(player['position'])
player_kit_number.append(player['shirtNo'])
match_players_df['playerId'] = player_ids
match_players_df['playerName'] = player_names
match_players_df['playerPos'] = player_pos
match_players_df['playerKitNumber'] = player_kit_number
# extracting passes
passes_df = events_df.loc[events_df['teamId'] ==
teamId].reset_index().drop('index', axis=1)
passes_df['playerId'] = passes_df['playerId'].astype('float').astype(
'Int64')
if 'playerName' in passes_df.columns:
passes_df = passes_df.drop(columns='playerName')
passes_df.dropna(subset=["playerId"], inplace=True)
passes_df.insert(
27,
column='playerName',
value=[team_players_dict[i] for i in list(passes_df['playerId'])])
if 'passRecipientId' in passes_df.columns:
passes_df = passes_df.drop(columns='passRecipientId')
passes_df = passes_df.drop(columns='passRecipientName')
passes_df.insert(28,
column='passRecipientId',
value=passes_df['playerId'].shift(-1))
passes_df.insert(29,
column='passRecipientName',
value=passes_df['playerName'].shift(-1))
passes_df.dropna(subset=["passRecipientName"], inplace=True)
passes_df = passes_df.loc[events_df['type'] == 'Pass', :].reset_index(
drop=True)
passes_df = passes_df.loc[events_df['outcomeType'] ==
'Successful', :].reset_index(drop=True)
index_names = passes_df.loc[passes_df['playerName'] ==
passes_df['passRecipientName']].index
passes_df.drop(index_names, inplace=True)
passes_df = passes_df.merge(match_players_df,
on=['playerId', 'playerName'],
how='left',
validate='m:1')
passes_df = passes_df.merge(match_players_df.rename(
{
'playerId': 'passRecipientId',
'playerName': 'passRecipientName'
},
axis='columns'),
on=['passRecipientId', 'passRecipientName'],
how='left',
validate='m:1',
suffixes=['', 'Receipt'])
passes_df = passes_df[passes_df['playerPos'] != 'Sub']
# getting team formation
formation = match_data[venue]['formations'][0]['formationName']
formation = '-'.join(formation)
# getting player average locations
location_formation = passes_df[['playerKitNumber', 'x', 'y']]
average_locs_and_count = location_formation.groupby('playerKitNumber').agg(
{
'x': ['mean'],
'y': ['mean', 'count']
})
average_locs_and_count.columns = ['x', 'y', 'count']
# getting separate dataframe for selected columns
passes_formation = passes_df[[
'id', 'playerKitNumber', 'playerKitNumberReceipt'
]].copy()
passes_formation['EPV'] = passes_df['EPV']
# getting dataframe for passes between players
passes_between = passes_formation.groupby(
['playerKitNumber', 'playerKitNumberReceipt']).agg({
'id': 'count',
'EPV': 'sum'
}).reset_index()
passes_between.rename({'id': 'pass_count'}, axis='columns', inplace=True)
passes_between = passes_between.merge(average_locs_and_count,
left_on='playerKitNumberReceipt',
right_index=True)
passes_between = passes_between.merge(average_locs_and_count,
left_on='playerKitNumber',
right_index=True,
suffixes=['', '_end'])
# filtering passes
pass_filter = int(passes_between['pass_count'].mean())
passes_between = passes_between.loc[
passes_between['pass_count'] > pass_filter]
# calculating the line width
passes_between[
'width'] = passes_between.pass_count / passes_between.pass_count.max(
) * max_line_width
passes_between = passes_between.reset_index(drop=True)
# setting color to make the lines more transparent when fewer passes are made
min_transparency = 0.3
color = np.array(to_rgba('white'))
color = np.tile(color, (len(passes_between), 1))
c_transparency = passes_between.pass_count / passes_between.pass_count.max(
)
c_transparency = (c_transparency *
(1 - min_transparency)) + min_transparency
color[:, 3] = c_transparency
passes_between['alpha'] = color.tolist()
# separating paired passes from normal passes
passes_between_threshold = 15
filtered_pair_df = []
pair_list = [
comb
for comb in combinations(passes_between['playerKitNumber'].unique(), 2)
]
for pair in pair_list:
df = passes_between[(
(passes_between['playerKitNumber'] == pair[0]) &
(passes_between['playerKitNumberReceipt'] == pair[1])) | (
(passes_between['playerKitNumber'] == pair[1]) &
(passes_between['playerKitNumberReceipt'] == pair[0]))]
if df.shape[0] == 2:
if (np.array(df.pass_count)[0] >= passes_between_threshold) and (
np.array(df.pass_count)[1] >= passes_between_threshold):
filtered_pair_df.append(df)
passes_between.drop(df.index, inplace=True)
if len(filtered_pair_df) > 0:
filtered_pair_df = pd.concat(filtered_pair_df).reset_index(drop=True)
passes_between = passes_between.reset_index(drop=True)
# plotting
pitch = Pitch(pitch_type='opta',
pitch_color='#171717',
line_color='#5c5c5c',
goal_type='box')
pitch.draw(ax=ax, constrained_layout=True, tight_layout=True)
average_locs_and_count['zorder'] = list(np.linspace(1, 5, 11))
for i in average_locs_and_count.index:
pitch.scatter(average_locs_and_count.loc[i, 'x'],
average_locs_and_count.loc[i, 'y'],
s=marker_size,
color=marker_color,
edgecolors=marker_edge_color,
linewidth=edgewidth,
alpha=1,
zorder=average_locs_and_count.loc[i, 'zorder'],
ax=ax)
for i in passes_between.index:
x = passes_between.loc[i, 'x']
y = passes_between.loc[i, 'y']
endX = passes_between.loc[i, 'x_end']
endY = passes_between.loc[i, 'y_end']
coordsA = "data"
coordsB = "data"
con = ConnectionPatch([endX, endY], [x, y],
coordsA,
coordsB,
arrowstyle="simple",
shrinkA=shrink,
shrinkB=shrink,
mutation_scale=passes_between.loc[i, 'width'] *
max_line_width,
color=passes_between.loc[i, 'alpha'])
ax.add_artist(con)
if len(filtered_pair_df) > 0:
for i in filtered_pair_df.index:
x = filtered_pair_df.loc[i, 'x']
y = filtered_pair_df.loc[i, 'y']
endX = filtered_pair_df.loc[i, 'x_end']
endY = filtered_pair_df.loc[i, 'y_end']
coordsA = "data"
coordsB = "data"
con = ConnectionPatch([endX, endY], [x, y],
coordsA,
coordsB,
arrowstyle="<|-|>",
shrinkA=shrink,
shrinkB=shrink,
mutation_scale=dh_arrow_width,
lw=filtered_pair_df.loc[i, 'width'] *
max_line_width / 5,
color=filtered_pair_df.loc[i, 'alpha'])
ax.add_artist(con)
for i in average_locs_and_count.index:
pitch.annotate(i,
xy=(average_locs_and_count.loc[i, 'x'],
average_locs_and_count.loc[i, 'y']),
family='DejaVu Sans',
c='white',
va='center',
ha='center',
zorder=average_locs_and_count.loc[i, 'zorder'],
size=kit_no_size,
weight='bold',
ax=ax)
ax.text(50,
104,
"{} (Mins 1-{})".format(team, sub_minute).upper(),
size=10,
fontweight='bold',
ha='center',
va='center')
ax.text(2, 3, '{}'.format(formation), size=9, c='grey')
"""
====================
Plot rotated markers
====================
This example shows how to plot rotated markers.
Warning: The rotation angle is in degrees and assumes the original marker is pointing upwards ↑.
If it's not you will have to modify the rotation degrees.
Rotates the marker in degrees, clockwise. 0 degrees is facing the direction of play (left to right).
In a horizontal pitch, 0 degrees is this way →, in a vertical pitch, 0 degrees is this way ↑
"""
from mplsoccer.pitch import Pitch
import matplotlib.pyplot as plt
from mplsoccer.scatterutils import arrowhead_marker
import numpy as np
plt.style.use('ggplot')
pitch = Pitch(figsize=(10, 8), axis=True, label=True)
fig, ax = pitch.draw()
n = 15
x = np.linspace(0, 120, n)
y = np.linspace(0, 80, n)
rotate = np.linspace(0, 360, n).round(0).astype(np.int32)
pitch.scatter(x, y, rotation_degrees=rotate, marker=arrowhead_marker, edgecolors='black', s=3500, ax=ax)
for i in range(n):
pitch.annotate(f'{rotate[i]}°', (x[i], y[i]), ha='center', va='center', fontsize=15, color='white', ax=ax)
def createPVFormationMap(match_data, events_df, team, color_palette,
markerstyle, markersize, markeredgewidth, labelsize,
labelcolor, ax):
# getting team id and venue
if match_data['home']['name'] == team:
teamId = match_data['home']['teamId']
venue = 'home'
else:
teamId = match_data['away']['teamId']
venue = 'away'
# getting opponent
if venue == 'home':
opponent = match_data['away']['name']
else:
opponent = match_data['home']['name']
# getting player dictionary
team_players_dict = {}
for player in match_data[venue]['players']:
team_players_dict[player['playerId']] = player['name']
# getting minute of first substitution
for i, row in events_df.iterrows():
if row['type'] == 'SubstitutionOn' and row['teamId'] == teamId:
sub_minute = str(row['minute'])
break
# getting players dataframe
match_players_df = pd.DataFrame()
player_names = []
player_ids = []
player_pos = []
player_kit_number = []
for player in match_data[venue]['players']:
player_names.append(player['name'])
player_ids.append(player['playerId'])
player_pos.append(player['position'])
player_kit_number.append(player['shirtNo'])
match_players_df['playerId'] = player_ids
match_players_df['playerName'] = player_names
match_players_df['playerPos'] = player_pos
match_players_df['playerKitNumber'] = player_kit_number
# extracting passes
passes_df = events_df.loc[events_df['teamId'] ==
teamId].reset_index().drop('index', axis=1)
passes_df['playerId'] = passes_df['playerId'].astype('float').astype(
'Int64')
if 'playerName' in passes_df.columns:
passes_df = passes_df.drop(columns='playerName')
passes_df.dropna(subset=["playerId"], inplace=True)
passes_df.insert(
27,
column='playerName',
value=[team_players_dict[i] for i in list(passes_df['playerId'])])
if 'passRecipientId' in passes_df.columns:
passes_df = passes_df.drop(columns='passRecipientId')
passes_df = passes_df.drop(columns='passRecipientName')
passes_df.insert(28,
column='passRecipientId',
value=passes_df['playerId'].shift(-1))
passes_df.insert(29,
column='passRecipientName',
value=passes_df['playerName'].shift(-1))
passes_df.dropna(subset=["passRecipientName"], inplace=True)
passes_df = passes_df.loc[events_df['type'] == 'Pass', :].reset_index(
drop=True)
passes_df = passes_df.loc[events_df['outcomeType'] ==
'Successful', :].reset_index(drop=True)
index_names = passes_df.loc[passes_df['playerName'] ==
passes_df['passRecipientName']].index
passes_df.drop(index_names, inplace=True)
passes_df = passes_df.merge(match_players_df,
on=['playerId', 'playerName'],
how='left',
validate='m:1')
passes_df = passes_df.merge(match_players_df.rename(
{
'playerId': 'passRecipientId',
'playerName': 'passRecipientName'
},
axis='columns'),
on=['passRecipientId', 'passRecipientName'],
how='left',
validate='m:1',
suffixes=['', 'Receipt'])
# passes_df = passes_df[passes_df['playerPos'] != 'Sub']
# Getting net possesion value for passes
netPVPassed = passes_df.groupby(['playerId',
'playerName'])['EPV'].sum().reset_index()
netPVReceived = passes_df.groupby(['passRecipientId', 'passRecipientName'
])['EPV'].sum().reset_index()
# Getting formation and player ids for first 11
formation = match_data[venue]['formations'][0]['formationName']
formation_positions = match_data[venue]['formations'][0][
'formationPositions']
playerIds = match_data[venue]['formations'][0]['playerIds'][:11]
# Getting all data in a dataframe
formation_data = []
for playerId, pos in zip(playerIds, formation_positions):
pl_dict = {'playerId': playerId}
pl_dict.update(pos)
formation_data.append(pl_dict)
formation_data = pd.DataFrame(formation_data)
formation_data['vertical'] = normalize(
formation_data['vertical'], {
'actual': {
'lower': 0,
'upper': 10
},
'desired': {
'lower': 10,
'upper': 110
}
})
formation_data['horizontal'] = normalize(formation_data['horizontal'], {
'actual': {
'lower': 0,
'upper': 10
},
'desired': {
'lower': 80,
'upper': 0
}
})
formation_data = netPVPassed.join(formation_data.set_index('playerId'),
on='playerId',
how='inner').reset_index(drop=True)
formation_data = formation_data.rename(columns={"EPV": "PV"})
# Plotting
pitch = Pitch(pitch_type='statsbomb',
pitch_color='#171717',
line_color='#5c5c5c',
goal_type='box')
pitch.draw(ax=ax, constrained_layout=True, tight_layout=True)
sns.scatterplot(x='vertical',
y='horizontal',
data=formation_data,
hue='PV',
s=markersize,
marker=markerstyle,
legend=False,
palette=color_palette,
linewidth=markeredgewidth,
ax=ax)
ax.text(2, 78, '{}'.format('-'.join(formation)), size=20, c='grey')
for index, row in formation_data.iterrows():
pitch.annotate(str(round(row.PV * 100, 2)) + '%',
xy=(row.vertical, row.horizontal),
c=labelcolor,
va='center',
ha='center',
size=labelsize,
zorder=2,
weight='bold',
ax=ax)
pitch.annotate(row.playerName,
xy=(row.vertical, row.horizontal + 5),
c=labelcolor,
va='center',
ha='center',
size=labelsize,
zorder=2,
weight='bold',
ax=ax)
arrows = pitch.lines(1.2 * pass_between.x,
.8 * pass_between.y,
1.2 * pass_between.x_end,
.8 * pass_between.y_end,
ax=ax,
lw=pass_between.pass_count * 1,
color='White',
zorder=1,
alpha=0.5)
nodes = pitch.scatter(1.2 * average_locations.x,
.8 * average_locations.y,
s=600,
color='#d3d3d3',
edgecolors='black',
linewidth=2.5,
alpha=1,
ax=ax)
for index, row in average_locations.iterrows():
pitch.annotate(int(row.passer),
xy=(row.x * 1.2, row.y * .8),
c='Black',
va='center',
ha='center',
size=13,
weight='bold',
ax=ax)
ax.set_title('Barcelona vs Valladolida', color='#dee6ea', fontsize=30)
ax.set_xlabel('@ibramallu')
def createPassNetworks(match_data, matches_df, events_df, team,
pitch_color, max_lw, marker_size, marker_color, marker_label='kit_no', marker_label_size=20):
"""
Parameters
----------
match_data : Data containing everything about the match
matches_df : DataFrame containing match data.
events_df : DataFrame containing event data for that match.
team : Name of the required team.
pitch_color : color of the pitch and figure.
max_lw : maximum line width of network lines.
marker_size : size of the circle markers.
marker_color : color of the circle markers.
Returns
-------
Pitch Plot.
"""
matchId = match_data['matchId']
# getting team id and venue
if match_data['home']['name'] == team:
teamId = match_data['home']['teamId']
venue = 'home'
else:
teamId = match_data['away']['teamId']
venue = 'away'
# getting opponent
if venue == 'home':
opponent = match_data['away']['name']
else:
opponent = match_data['home']['name']
team_players_dict = {}
for player in matches_df[venue][matchId]['players']:
team_players_dict[player['playerId']] = player['name']
team_playerskitno_dict = {}
for player in matches_df[venue][matchId]['players']:
team_playerskitno_dict[player['shirtNo']] = player['name']
passes_df = events_df.loc[[row['displayName'] == 'Pass' for row in list(events_df['type'])]].reset_index(drop=True)
passes_df = passes_df.loc[passes_df['teamId'] == teamId].reset_index().drop('index', axis=1)
passes_df = passes_df.loc[[row['displayName'] == 'Successful' for row in list(passes_df['outcomeType'])]].reset_index(drop=True)
passes_df.insert(27, column='playerName', value=[team_players_dict[i] for i in list(passes_df['playerId'])])
passes_df.insert(28, column='passRecipientId', value=passes_df['playerId'].shift(-1))
passes_df.insert(29, column='passRecipientName', value=passes_df['playerName'].shift(-1))
passes_df.dropna(subset=["passRecipientName"], inplace=True)
match_players_df = pd.DataFrame()
player_names = []
player_ids = []
player_pos = []
player_kit_number = []
for player in matches_df[venue][matchId]['players']:
player_names.append(player['name'])
player_ids.append(player['playerId'])
player_pos.append(player['position'])
player_kit_number.append(player['shirtNo'])
match_players_df['playerId'] = player_ids
match_players_df['playerName'] = player_names
match_players_df['playerPos'] = player_pos
match_players_df['playerKitNumber'] = player_kit_number
passes_df = passes_df.merge(match_players_df, on=['playerId', 'playerName'], how='left', validate='m:1')
passes_df = passes_df.merge(match_players_df.rename({'playerId': 'passRecipientId', 'playerName':'passRecipientName'},
axis='columns'), on=['passRecipientId', 'passRecipientName'],
how='left', validate='m:1', suffixes=['', 'Receipt'])
passes_df = passes_df[passes_df['playerPos'] != 'Sub']
passes_formation = passes_df[['id', 'playerKitNumber', 'playerKitNumberReceipt']].copy()
location_formation = passes_df[['playerKitNumber', 'x', 'y']]
average_locs_and_count = location_formation.groupby('playerKitNumber').agg({'x': ['mean'], 'y': ['mean', 'count']})
average_locs_and_count.columns = ['x', 'y', 'count']
passes_formation['kitNo_max'] = passes_formation[['playerKitNumber',
'playerKitNumberReceipt']].max(axis='columns')
passes_formation['kitNo_min'] = passes_formation[['playerKitNumber',
'playerKitNumberReceipt']].min(axis='columns')
passes_between = passes_formation.groupby(['kitNo_max', 'kitNo_min']).id.count().reset_index()
passes_between.rename({'id': 'pass_count'}, axis='columns', inplace=True)
# add on the location of each player so we have the start and end positions of the lines
passes_between = passes_between.merge(average_locs_and_count, left_on='kitNo_min', right_index=True)
passes_between = passes_between.merge(average_locs_and_count, left_on='kitNo_max', right_index=True,
suffixes=['', '_end'])
##############################################################################
# Calculate the line width and marker sizes relative to the largest counts
max_line_width = max_lw
passes_between['width'] = passes_between.pass_count / passes_between.pass_count.max() * max_line_width
#average_locs_and_count['marker_size'] = (average_locs_and_count['count']
# / average_locs_and_count['count'].max() * max_marker_size)
##############################################################################
# Set color to make the lines more transparent when fewer passes are made
min_transparency = 0.3
color = np.array(to_rgba('white'))
color = np.tile(color, (len(passes_between), 1))
c_transparency = passes_between.pass_count / passes_between.pass_count.max()
c_transparency = (c_transparency * (1 - min_transparency)) + min_transparency
color[:, 3] = c_transparency
# Plotting name on markers
if marker_label == 'name':
average_locs_and_count.index = average_locs_and_count.index.map(team_playerskitno_dict)
##############################################################################
# Plotting
pitch = Pitch(pitch_type='statsbomb', orientation='horizontal',
pitch_color=pitch_color, line_color='#c7d5cc', figsize=(16, 11),
constrained_layout=True, tight_layout=False)
fig, ax = pitch.draw()
pitch.lines(passes_between.x/100*120, 80-passes_between.y/100*80,
passes_between.x_end/100*120, 80-passes_between.y_end/100*80, lw=passes_between.width,
color=color, zorder=1, ax=ax)
pitch.scatter(average_locs_and_count.x/100*120, 80-average_locs_and_count.y/100*80, s=marker_size,
color=marker_color, edgecolors='black', linewidth=1, alpha=1, ax=ax)
for index, row in average_locs_and_count.iterrows():
pitch.annotate(row.name, xy=(row.x/100*120, 80-row.y/100*80), c='white', va='center', ha='center', size=marker_label_size, weight='bold', ax=ax)
ax.set_title("{} Pass Network vs {}".format(team, opponent), size=15, y=0.97, color='#c7d5cc')
fig.set_facecolor(pitch_color)
def createAttPassNetworks(match_data, matches_df, events_df, team, pitch_color, max_lw,
marker_size, marker_color, marker_label='kit_no', marker_label_size=20):
"""
Parameters
----------
match_data : Data containing everything about the match
matches_df : DataFrame containing match data.
events_df : DataFrame containing event data for that match.
team : Name of the required team.
pitch_color : color of the pitch and figure.
max_lw : maximum line width of network lines.
marker_size : size of the circle markers.
marker_color : color of the circle markers.
Returns
-------
Pitch Plot.
"""
# getting match_id
matchId = match_data['matchId']
# getting team id and venue
if match_data['home']['name'] == team:
teamId = match_data['home']['teamId']
venue = 'home'
else:
teamId = match_data['away']['teamId']
venue = 'away'
# getting opponent
if venue == 'home':
opponent = match_data['away']['name']
else:
opponent = match_data['home']['name']
# getting player dictionary
team_players_dict = {}
for player in match_data[venue]['players']:
team_players_dict[player['playerId']] = player['name']
# getting player dictionary with kit no
team_playerskitno_dict = {}
for player in matches_df[venue][matchId]['players']:
team_playerskitno_dict[player['shirtNo']] = player['name']
# getting minute of first substitution
for i,row in events_df.iterrows():
if row['type']['displayName'] == 'SubstitutionOn' and row['teamId'] == teamId:
sub_minute = str(row['minute'])
break
# getting players dataframe
match_players_df = pd.DataFrame()
player_names = []
player_ids = []
player_pos = []
player_kit_number = []
for player in match_data[venue]['players']:
player_names.append(player['name'])
player_ids.append(player['playerId'])
player_pos.append(player['position'])
player_kit_number.append(player['shirtNo'])
match_players_df['playerId'] = player_ids
match_players_df['playerName'] = player_names
match_players_df['playerPos'] = player_pos
match_players_df['playerKitNumber'] = player_kit_number
# extracting passes
passes_df = events_df.loc[events_df['teamId'] == teamId].reset_index().drop('index', axis=1)
passes_df.dropna(subset=["playerId"], inplace=True)
passes_df.insert(27, column='playerName', value=[team_players_dict[i] for i in list(passes_df['playerId'])])
passes_df.insert(28, column='passRecipientId', value=passes_df['playerId'].shift(-1))
passes_df.insert(29, column='passRecipientName', value=passes_df['playerName'].shift(-1))
passes_df.dropna(subset=["passRecipientName"], inplace=True)
passes_df = passes_df.loc[[row['displayName'] == 'Pass' for row in list(passes_df['type'])]].reset_index(drop=True)
passes_df = passes_df.loc[[row['displayName'] == 'Successful' for row in list(passes_df['outcomeType'])]].reset_index(drop=True)
index_names = passes_df.loc[passes_df['playerName']==passes_df['passRecipientName']].index
passes_df.drop(index_names, inplace=True)
passes_df = passes_df.merge(match_players_df, on=['playerId', 'playerName'], how='left', validate='m:1')
passes_df = passes_df.merge(match_players_df.rename({'playerId': 'passRecipientId', 'playerName':'passRecipientName'},
axis='columns'), on=['passRecipientId', 'passRecipientName'],
how='left', validate='m:1', suffixes=['', 'Receipt'])
passes_df = passes_df[passes_df['playerPos'] != 'Sub']
# getting team formation
formation = match_data[venue]['formations'][0]['formationName']
formation = '-'.join(formation)
# getting player average locations
location_formation = passes_df[['playerKitNumber', 'x', 'y']]
average_locs_and_count = location_formation.groupby('playerKitNumber').agg({'x': ['mean'], 'y': ['mean', 'count']})
average_locs_and_count.columns = ['x', 'y', 'count']
# filtering progressive passes
passes_df = passes_df.loc[passes_df['EPV_difference'] > 0]
# getting separate dataframe for selected columns
passes_formation = passes_df[['id', 'playerKitNumber', 'playerKitNumberReceipt']].copy()
passes_formation['EPV'] = passes_df['EPV_difference']
# getting dataframe for passes between players
passes_between = passes_formation.groupby(['playerKitNumber', 'playerKitNumberReceipt']).agg({ 'id' : 'count', 'EPV' : 'sum'}).reset_index()
passes_between.rename({'id': 'pass_count'}, axis='columns', inplace=True)
passes_between = passes_between.merge(average_locs_and_count, left_on='playerKitNumberReceipt', right_index=True)
passes_between = passes_between.merge(average_locs_and_count, left_on='playerKitNumber', right_index=True,
suffixes=['', '_end'])
# filtering passes
pass_filter = int(passes_between['pass_count'].mean())
passes_between = passes_between.loc[passes_between['pass_count'] > pass_filter*2]
# calculating the line width
max_line_width = max_lw
passes_between['width'] = passes_between.pass_count / passes_between.pass_count.max() * max_line_width
passes_between = passes_between.reset_index(drop=True)
# setting color to make the lines more transparent when fewer passes are made
min_transparency = 0.3
color = np.array(to_rgba('white'))
color = np.tile(color, (len(passes_between), 1))
c_transparency = passes_between.EPV / passes_between.EPV.max()
c_transparency = (c_transparency * (1 - min_transparency)) + min_transparency
color[:, 3] = c_transparency
passes_between['alpha'] = color.tolist()
# Plotting name on markers
if marker_label == 'name':
average_locs_and_count.index = average_locs_and_count.index.map(team_playerskitno_dict)
# plotting
pitch = Pitch(pitch_type='statsbomb', orientation='horizontal',
pitch_color=pitch_color, line_color='#c7d5cc', figsize=(16, 11),
constrained_layout=True, tight_layout=False)
fig, ax = pitch.draw()
average_locs_and_count['zorder'] = list(np.linspace(2,6,11, dtype='int'))
pitch.lines(passes_between.x/100*120, 80-passes_between.y/100*80,
passes_between.x_end/100*120, 80-passes_between.y_end/100*80, lw=passes_between.width,
color=color, zorder=1, ax=ax)
for index, row in average_locs_and_count.iterrows():
pitch.scatter(row.x/100*120, 80-row.y/100*80, s=marker_size,
color=marker_color, edgecolors='black', linewidth=1,
alpha=1, zorder=row.zorder, ax=ax)
for index, row in average_locs_and_count.iterrows():
pitch.annotate(row.name, xy=(row.x/100*120, 80-row.y/100*80), family='DejaVu Sans', c='white',
va='center', ha='center', zorder=row.zorder, size=marker_label_size, weight='bold', ax=ax)
ax.set_title("{} Progressive Pass Network vs {}".format(team, opponent), size=15, y=0.97, color='#c7d5cc')
fig.set_facecolor(pitch_color)
def passmap(Df, teamid, teamname, min1, max1):
pitch = Pitch(pitch_type='statsbomb',
orientation='vertical',
pitch_color='#000000',
line_color='#a9a9a9',
constrained_layout=True,
tight_layout=False,
linewidth=0.5)
fig, ax = pitch.draw()
df = Df.copy()
df = df[(df.expandedMinute >= min1) & (df.expandedMinute <= max1)]
allplayers = df[(df.teamId == teamid) & (df.name.notna())].name.tolist()
playersubbedoff = df[(df.type_displayName == 'SubstitutionOff')
& (df.teamId == teamid)]['name'].tolist()
timeoff = df[(df.type_displayName == 'SubstitutionOff')
& (df.teamId == teamid)]['expandedMinute'].tolist()
playersubbedon = df[(df.type_displayName == 'SubstitutionOn')
& (df.teamId == teamid)]['name'].tolist()
timeon = df[(df.type_displayName == 'SubstitutionOn')
& (df.teamId == teamid)]['expandedMinute'].tolist()
majoritylist = []
minoritylist = []
for i in range(len(timeon)):
if ((timeon[i] >= min1) & (timeon[i] <= max1)):
player1min = timeon[i] - min1
player2min = max1 - timeon[i]
if (player1min >= player2min):
majoritylist.append(playersubbedoff[i])
minoritylist.append(playersubbedon[i])
else:
majoritylist.append(playersubbedon[i])
minoritylist.append(playersubbedoff[i])
players = list(set(allplayers) - set(minoritylist))
#return players
shirtNo = []
for p in players:
shirtNo.append(int(df[df.name == p]['shirtNo'].values[0]))
passes_df = df.query(
"(type_displayName=='Pass')&(name in @players)&(receiver in @players)&\
(outcomeType_displayName == 'Successful')&(teamId==@teamid)")
#passes_df.insert(29, column='passRecipientName', value=passes_df['name'].shift(-1))
passes_df.dropna(subset=["receiver"], inplace=True)
#passes_df['passer'] = passes_df['playerId']
#passes_df['recipient'] = passes_df['passer'].shift(-1)
passes_df = passes_df[passes_df.columns[~passes_df.isnull().all()]]
passes_df['playerKitNumber'] = passes_df['shirtNo'].fillna(0).astype(
np.int)
passes_df['playerKitNumberReceipt'] = passes_df['shirtNo'].shift(
-1).fillna(0).astype(np.int)
passer_avg = passes_df.groupby('playerKitNumber').agg({
'x': ['median'],
'y': ['median', 'count'],
'epv_value': ['sum']
})
passer_avg.columns = ['x', 'y', 'count', 'epv']
passer_avg.index = passer_avg.index.astype(int)
passes_formation = passes_df[[
'id', 'playerKitNumber', 'playerKitNumberReceipt'
]].copy()
passes_formation['kitNo_max'] = passes_formation[[
'playerKitNumber', 'playerKitNumberReceipt'
]].max(axis='columns')
passes_formation['kitNo_min'] = passes_formation[[
'playerKitNumber', 'playerKitNumberReceipt'
]].min(axis='columns')
passes_between = passes_formation.groupby(['kitNo_max', 'kitNo_min'
]).id.count().reset_index()
passes_between.rename({'id': 'pass_count'}, axis='columns', inplace=True)
# add on the location of each player so we have the start and end positions of the lines
passes_between = passes_between.merge(passer_avg,
left_on='kitNo_min',
right_index=True)
passes_between = passes_between.merge(passer_avg,
left_on='kitNo_max',
right_index=True,
suffixes=['', '_end'])
'''
#Between Passer and Recipient
passes_between = passes_df.groupby(['passer', 'recipient']).id.count().reset_index()
passes_between.rename({'id': 'pass_count'}, axis='columns', inplace=True)
passes_between = passes_between.merge(passer_avg, left_on='passer', right_index=True)
passes_between = passes_between.merge(passer_avg, left_on='recipient', right_index=True,
suffixes=['', '_end'])
'''
#Minimum No. of Passes
passes_between = passes_between.loc[(passes_between['pass_count'] >= 3)]
#Scaling for StatsBomb
passes_between['x'] = passes_between['x'] * 1.2
passes_between['y'] = passes_between['y'] * 0.8
passer_avg['x'] = passer_avg['x'] * 1.2
passer_avg['y'] = passer_avg['y'] * 0.8
passes_between['x_end'] = passes_between['x_end'] * 1.2
passes_between['y_end'] = passes_between['y_end'] * 0.8
#Width Variable
yo = passes_between.pass_count / passes_between.pass_count.max()
b = 1
min_transparency = 0.3
color = np.array(to_rgba('#00bfff'))
color = np.tile(color, (len(passes_between), 1))
c_transparency = passes_between.pass_count / passes_between.pass_count.max(
)
c_transparency = (c_transparency *
(1 - min_transparency)) + min_transparency
color[:, 3] = c_transparency
a = plt.scatter(passer_avg.y,
passer_avg.x,
s=100,
c="#111111",
facecolor='none',
lw=1,
cmap="winter",
alpha=1,
zorder=2,
vmin=0,
vmax=0.6,
marker='h')
c = plt.scatter(passer_avg.y,
passer_avg.x,
s=60,
c='#FF0000',
alpha=1,
zorder=3,
marker='h')
pitch.arrows(passes_between.x,
passes_between.y,
passes_between.x_end,
passes_between.y_end,
color=color,
ax=ax,
zorder=1,
width=1.5)
cbar = plt.colorbar(a,
orientation="horizontal",
shrink=0.3,
pad=0,
ticks=[0, 0.2, 0.4, 0.6])
cbar.set_label('Expected Possession Value (EPV)', color='#a9a9a9', size=6)
cbar.outline.set_edgecolor('#a9a9a9')
cbar.ax.xaxis.set_tick_params(color='#a9a9a9')
cbar.ax.xaxis.set_tick_params(labelcolor='#a9a9a9')
cbar.ax.tick_params(labelsize=5)
plt.gca().invert_xaxis()
for index, row in passer_avg.iterrows():
pitch.annotate(row.name,
xy=(row.x, row.y),
c='#a9a9a9',
va='center',
ha='center',
size=5,
ax=ax)
plt.text(
79,
2,
"Positions = Median Location of Successful Passes\nArrows = Pass Direction\nTransparency = Frequency of Combination\nMinimum of 3 Passes ",
color='#a9a9a9',
fontsize=5,
alpha=0.5,
zorder=1)
plt.text(80, 122, "Minutes 45-90", color='#a9a9a9', fontsize=5)
plt.text(18, 122, "@nikhilrajesh231", color='#a9a9a9', fontsize=5)
ax.set_title("Barcelona PV Pass Network\n5-2 vs Getafe (H)",
fontsize=8,
color="#a9a9a9",
fontweight='bold',
y=1.01)
fig = plt.savefig('pn2.png',
bbox_inches="tight",
facecolor="#000000",
dpi=600)
return fig