def plot_individual_cluster_events(rows, cols, events_df, cluster_labels, sample_size = 5, pitch_length=120, pitch_width=80, pitch_theme = 'dark', line_colour='white', ax_colour = '#303030', event_colour='royalblue', figsize=(10, 16)):
'''
Creates figure and axes grid using specified rows x columns.
Plots each cluster of StatsBomb event data on a separate pitch with specified number of sample events.
Parameters:
rows (integer): number of rows in axes grid
cols (integer): number of columns in axes grid
events_df (dataframe): event dataframe with x and y vertical locations
cluster_labels (list): list of labels assigned to each respective event
sample_size (integer): number of sample events to plot on each axes
pitch_length (integer): length of pitch in yards
pitch_width (integer): width of pitch in yards
metric (string): specify distance metric, yards (or metres - not yet available)
alpha (numeric): transparency of events, 0-1
event_colour (string): colour of events
pitch_theme (string): specify 'light' or 'dark' to auto set pitch and line colours
pitch_line_colour (string): specify colour for pitch lines
ax_colour (string): specify colour for axes background colour
figsize (tuple): specify (width, height) of figure
'''
import matplotlib.pyplot as plt
import collections
fig, axs = plt.subplots(rows,cols, sharex = True, sharey = True, figsize = figsize)
if pitch_theme == 'dark':
fig.patch.set_facecolor('#303030')
cluster_freq = collections.Counter(cluster_labels)
cluster_sorted = list(zip(*cluster_freq.most_common()))[0]
cluster_freq_sorted = list(zip(*cluster_freq.most_common()))[1]
for ax, cluster, event_count in zip(axs.flat, cluster_sorted, cluster_freq_sorted):
cluster_events = events_df[cluster_labels == cluster].sample(sample_size)
createVerticalPitch(length=pitch_length, width=pitch_width, metric='yards', pitch_theme = pitch_theme, linecolor=line_colour, ax_colour = ax_colour, figsize = figsize, figax = (fig, ax))
plot_sb_events(cluster_events, figax = (fig, ax), pitch_theme=pitch_theme, event_colour=event_colour)
ax.set_title("Cluster " + str(cluster+1) + " - (" + str(event_count) + ")",
fontdict = dict(fontweight='bold',
color='white'))
ax.axis('off')
plt.subplots_adjust()
plt.tight_layout()
return fig, axs
def plot_sb_event_location(events_df, pitch_length = 120, pitch_width = 80, metric = 'yards', alpha = 0.7, event_colour = 'royalblue', pitch_theme='light', pitch_line_colour = 'black', ax_colour = 'white', figsize=(5, 10), figax=None):
'''
Plots StatsBomb event data on a vertical pitch using transformed vertical locations.
Parameters:
events_df (dataframe): event dataframe with x and y vertical locations
pitch_length (integer): length of pitch in yards
pitch_width (integer): width of pitch in yards
metric (string): specify distance metric, yards (or metres - not yet available)
alpha (numeric): transparency of events, 0-1
event_colour (string): colour of events
pitch_theme (string): specify 'light' or 'dark' to auto set pitch and line colours
pitch_line_colour (string): specify colour for pitch lines
ax_colour (string): specify colour for axes background colour
figsize (tuple): specify (width, height) of figure
figax (tuple): specify previous (fig, ax) to start from
'''
fig,ax = createVerticalPitch(length=pitch_length, width=pitch_width, metric=metric, pitch_theme = pitch_theme, linecolor=pitch_line_colour, ax_colour = ax_colour, figsize = figsize, figax = figax)
for index, event in events_df.iterrows():
x = event['vertical_location_x']
y = event['vertical_location_y']
ax.plot(x, y, 'ro', alpha = alpha)
return fig,ax
def plot_histogram_ratio_pitch(events_1, events_2, pitch_length=120, pitch_width=80, metric='yards', line_colour='black', nbins=6, grid_colour_map='RdBu', figsize=(5, 10)):
'''
Calculate and plot the ratio of two 2D histograms with specified number of bins.
Parameters:
events_1 (dataframe): event dataframe with x and y vertical locations
events_2 (dataframe): event dataframe with x and y vertical locations
pitch_length (integer): length of pitch in yards
pitch_width (integer): width of pitch in yards
metric (string): specify distance metric, yards (or metres - not yet available)
pitch_line_colour (string): specify colour for pitch lines
nbins (integer): number of bins for 2D histogram
grid_colour_map (string): Matplotlib colour map
figsize (tuple): specify (width, height) of figure
'''
x1, y1 = events_1['vertical_location_x'], events_1['vertical_location_y']
x2, y2 = events_2['vertical_location_x'], events_2['vertical_location_y']
h1, xedges, yedges = np.histogram2d(x1, y1, bins=nbins)
h2, xedges, yedges = np.histogram2d(x2, y2, bins=nbins)
h = -1 * (h1 / h2)
fig, ax = createVerticalPitch(pitch_length, pitch_width, metric, linecolor=line_colour, figsize = figsize)
ax.pcolorfast(xedges, yedges, h.T, cmap=grid_colour_map)
return fig, ax
def plot_sb_events_clusters(events_df, clusters=4, pitch_length = 120, pitch_width = 80, metric = 'yards', pitch_theme = 'light', line_colour = 'black', ax_colour = 'white', alpha = 0.7, figsize = (5, 10), figax=None):
'''
Applies k-means clustering to events with given number of clusters.
Then plots StatsBomb event data on a vertical pitch using transformed vertical start and end locations with random colours per cluster.
Parameters:
events_df (dataframe): event dataframe with x and y vertical locations
clusters (integer): number of clusters to use for k-means
pitch_length (integer): length of pitch in yards
pitch_width (integer): width of pitch in yards
metric (string): specify distance metric, yards (or metres - not yet available)
alpha (numeric): transparency of events, 0-1
event_colour (string): colour of events
pitch_theme (string): specify 'light' or 'dark' to auto set pitch and line colours
pitch_line_colour (string): specify colour for pitch lines
ax_colour (string): specify colour for axes background colour
figsize (tuple): specify (width, height) of figure
figax (tuple): specify previous (fig, ax) to start from
'''
from ClusterEval import kmeans_cluster, cluster_colour_map
cluster_labels = kmeans_cluster(events_df, clusters)
label_colour = cluster_colour_map(cluster_labels, clusters)
fig,ax = createVerticalPitch(length=pitch_length, width=pitch_width, metric=metric, pitch_theme = pitch_theme, linecolor=line_colour, ax_colour = ax_colour, figsize = figsize, figax = figax)
cluster_colour=0
for index, event in events_df.iterrows():
x_start = event['vertical_location_x']
y_start = event['vertical_location_y']
x_end = event['vertical_end_location_x']
y_end = event['vertical_end_location_y']
ax.annotate("", xy=(x_end, y_end), xytext = (x_start, y_start), alpha = alpha,
arrowprops = dict(alpha=alpha, arrowstyle="->", color = label_colour[cluster_colour]))
cluster_colour=cluster_colour+1
return fig, ax, cluster_labels
def plot_sb_event_grid_density_pitch(events_df, pitch_length = 120, pitch_width = 80, metric = 'yards', pitch_line_colour='black',spacing = 0.005, nbins = 6, grid_colour_map = 'Reds', bar_colour = 'Red', figsize=(5, 10)):
'''
Plot a 2D histogram of event locations with marginal density plots both vertically and horizontally.
Parameters:
events_df (dataframe): event dataframe with x and y vertical locations
pitch_length (integer): length of pitch in yards
pitch_width (integer): width of pitch in yards
metric (string): specify distance metric, yards (or metres - not yet available)
pitch_line_colour (string): specify colour for pitch lines
nbins (integer): number of bins for 2D histogram
grid_colour_map (string): Matplotlib colour map
bar_colour (string): colour of density plot bars
figsize (tuple): specify (width, height) of figure
'''
x = events_df['vertical_location_x']
y = events_df['vertical_location_y']
fig, ax = createVerticalPitch(pitch_length, pitch_width, metric, linecolor=pitch_line_colour, figsize = figsize)
ax_pos = ax.get_position()
left, width = ax_pos.x0, ax_pos.x1 - ax_pos.x0
bottom, height = ax_pos.y0, ax_pos.y1 - ax_pos.y0
rect_histx = [left, bottom + height + spacing, width, 0.1]
rect_histy = [left + width + spacing, bottom, 0.2, height]
ax_histx = fig.add_axes(rect_histx, sharex=ax)
ax_histy = fig.add_axes(rect_histy, sharey=ax)
ax, ax_x, ax_y = marginal_dist_grid(x, y, ax, ax_histx, ax_histy, nbins = nbins, grid_colour_map = grid_colour_map, bar_colour = bar_colour)
return fig, ax, ax_x, ax_y