# get data for a Sevilla versus Barcelona match with a high amount of shots kwargs = { 'related_event_df': False, 'shot_freeze_frame_df': False, 'tactics_lineup_df': False, 'warn': False } df = read_event(f'{EVENT_SLUG}/9860.json', **kwargs)['event'] # setup the mplsoccer StatsBomb Pitches # note not much padding around the pitch so the marginal axis are tight to the pitch # if you are using a different goal type you will need to increase the padding to see the goals pitch = Pitch(pad_top=0.05, pad_right=0.05, pad_bottom=0.05, pad_left=0.05, line_zorder=2) vertical_pitch = VerticalPitch(half=True, pad_top=0.05, pad_right=0.05, pad_bottom=0.05, pad_left=0.05, line_zorder=2) # setup a mplsoccer FontManager to download google fonts (Roboto-Regular / SigmarOne-Regular) fm = FontManager() fm_rubik = FontManager( ('https://github.com/google/fonts/blob/main/ofl/rubikmonoone/' 'RubikMonoOne-Regular.ttf?raw=true'))
############################################################################## # Boolean mask for filtering the dataset by team team1, team2 = df.team_name.unique() mask_team1 = (df.type_name == 'Pass') & (df.team_name == team1) ############################################################################## # Filter dataset to only include one teams passes and get boolean mask for the completed passes df_pass = df.loc[mask_team1, ['x', 'y', 'end_x', 'end_y', 'outcome_name']] mask_complete = df_pass.outcome_name.isnull() ############################################################################## # Setup the pitch and number of bins pitch = Pitch(pitch_type='statsbomb', line_zorder=2, line_color='#c7d5cc', pitch_color='#22312b') bins = (6, 4) ############################################################################## # Plotting using a single color and length fig, ax = pitch.draw(figsize=(16, 11), constrained_layout=True, tight_layout=False) # plot the heatmap - darker colors = more passes originating from that square bs_heatmap = pitch.bin_statistic(df_pass.x, df_pass.y, statistic='count', bins=bins) hm = pitch.heatmap(bs_heatmap, ax=ax, cmap='Blues') # plot the pass flow map with a single color ('black') and length of the arrow (5)
############################################################################## # Rotated markers # --------------- # I also included a method for rotating markers in mplsoccer. # # 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 ↑ # # We are going to plot pass data as an arrowhead marker with the # arrow facing in the direction of the pass # The marker size is going to relate to the pass distance, # so larger markers mean the pass was longer. pitch = Pitch() fig, ax = pitch.draw(figsize=(14, 12)) angle, distance = pitch.calculate_angle_and_distance(df_pass_barca.x, df_pass_barca.y, df_pass_barca.end_x, df_pass_barca.end_y, standardized=False, degrees=True) sc = pitch.scatter( df_pass_barca.x, df_pass_barca.y, rotation_degrees=angle, c='#b94b75', # color for scatter in hex format edgecolors='#383838', alpha=0.9, s=(distance / distance.max()) * 900,
# Get the StatsBomb logo and Fonts LOGO_URL = 'https://raw.githubusercontent.com/statsbomb/open-data/master/img/statsbomb-logo.jpg' sb_logo = Image.open(urlopen(LOGO_URL)) # a FontManager object for using a google font (default Robotto) fm = FontManager() # path effects path_eff = [path_effects.Stroke(linewidth=3, foreground='black'), path_effects.Normal()] ############################################################################## # Plot the percentages # setup a mplsoccer pitch pitch = Pitch(line_zorder=2, line_color='black', pad_top=20) # mplsoccer calculates the binned statistics usually from raw locations, such as pressure events # for this example we will create a binned statistic dividing # the pitch into thirds for one point (0, 0) # we will fill this in a loop later with each team's statistics from the dataframe bin_statistic = pitch.bin_statistic([0], [0], statistic='count', bins=(3, 1)) GRID_HEIGHT = 0.8 CBAR_WIDTH = 0.03 fig, axs = pitch.grid(nrows=4, ncols=5, figheight=20, # leaves some space on the right hand side for the colorbar grid_width=0.88, left=0.025, endnote_height=0.06, endnote_space=0, # Turn off the endnote/title axis. I usually do this after # I am happy with the chart layout and text placement
df_total = pd.DataFrame(df[pressure_cols].sum()) df_total.columns = ['total'] df_total = df_total.T df_total = df_total.divide(df_total.sum(axis=1), axis=0) * 100 ############################################################################## # Calculate the percentages for each team and sort so that the teams which press higher are last df[pressure_cols] = df[pressure_cols].divide(df[pressure_cols].sum(axis=1), axis=0) * 100. df.sort_values(['Att 3rd', 'Def 3rd'], ascending=[True, False], inplace=True) ############################################################################## # Plot the percentages # setup a mplsoccer pitch pitch = Pitch(line_zorder=2, line_color='black') # mplsoccer calculates the binned statistics usually from raw locations, such as pressure events # for this example we will create a binned statistic dividing # the pitch into thirds for one point (0, 0) # we will fill this in a loop later with each team's statistics from the dataframe bin_statistic = pitch.bin_statistic([0], [0], statistic='count', bins=(3, 1)) # Plot fig, axes = pitch.draw(figsize=(16, 9), ncols=5, nrows=4, tight_layout=False, constrained_layout=True) axes = axes.ravel() teams = df['Squad'].values
'related_event_df': False, 'shot_freeze_frame_df': False, 'tactics_lineup_df': False, 'warn': False } df = pd.concat([ read_event(f'{EVENT_SLUG}/{file}', **kwargs)['event'] for file in match_files ]) # filter chelsea pressure events mask_chelsea_pressure = (df.team_name == 'Chelsea FCW') & (df.type_name == 'Pressure') df = df.loc[mask_chelsea_pressure, ['x', 'y']] ############################################################################## # Plot the heatmaps # setup pitch pitch = Pitch(pitch_type='statsbomb', line_zorder=2, line_color='white') # draw fig, ax = pitch.draw(figsize=(16, 9)) bin_statistic = pitch.bin_statistic(df.x, df.y, statistic='count', bins=(25, 25)) bin_statistic['statistic'] = gaussian_filter(bin_statistic['statistic'], 1) pcm = pitch.heatmap(bin_statistic, ax=ax, cmap='hot', edgecolors='#22312b') cbar = fig.colorbar(pcm, ax=ax) TITLE_STR = 'Location of pressure events - 3 home games for Chelsea FC Women' title = fig.suptitle(TITLE_STR, x=0.4, y=0.98, fontsize=23)
kwargs = { 'related_event_df': False, 'shot_freeze_frame_df': False, 'tactics_lineup_df': False, 'warn': False } df_false9 = read_event(f'{EVENT_SLUG}/69249.json', **kwargs)['event'] df_before_false9 = read_event(f'{EVENT_SLUG}/69251.json', **kwargs)['event'] # filter messi's actions (starting positions) df_false9 = df_false9.loc[df_false9.player_id == 5503, ['x', 'y']] df_before_false9 = df_before_false9.loc[df_before_false9.player_id == 5503, ['x', 'y']] ############################################################################## # plotting pitch = Pitch(pitch_type='statsbomb', pitch_color='#22312b', stripe=False, line_zorder=2) fig, ax = pitch.draw( figsize=(16, 9), nrows=1, ncols=2, ) pitch.hexbin(df_before_false9.x, df_before_false9.y, ax=ax[0], cmap='Blues') pitch.hexbin(df_false9.x, df_false9.y, ax=ax[1], cmap='Blues') TITLE_STR1 = 'Messi in the game directly before \n playing in the false 9 role' TITLE_STR2 = 'The first Game Messi \nplayed in the false 9 role' title1 = ax[0].set_title(TITLE_STR1, fontsize=25, pad=20) title2 = ax[1].set_title(TITLE_STR2, fontsize=25, pad=20)
fig = plt.figure(figsize=FIGSIZE) fm_rubik = FontManager(('https://github.com/google/fonts/blob/main/ofl/rubikmonoone/' 'RubikMonoOne-Regular.ttf?raw=true')) # layout specifications PAD = 1 pitch_spec = {'pad_left': PAD, 'pad_right': PAD, 'pad_bottom': PAD, 'pad_top': PAD, 'pitch_color': 'None'} pitch_width, pitch_length = 80, 105 pitch_width3, pitch_length3 = 60, 105 pitch_length4, pitch_width4 = 120, 68 pitch_length6, pitch_width6 = 85, 68 # define pitches (top left, top middle, top right, bottom left, bottom middle, bottom right) pitch1 = Pitch(pitch_type='custom', pitch_width=pitch_width, pitch_length=pitch_length, line_color='#b94e45', **pitch_spec) pitch2 = Pitch(pitch_type='statsbomb', **pitch_spec) pitch3 = Pitch(pitch_type='custom', pitch_width=pitch_width3, pitch_length=pitch_length3, line_color='#56ae6c', **pitch_spec) pitch4 = VerticalPitch(pitch_type='custom', pitch_length=pitch_length4, pitch_width=pitch_width4, line_color='#bc7d39', **pitch_spec) pitch5 = VerticalPitch(pitch_type='statsbomb', **pitch_spec) pitch6 = VerticalPitch(pitch_type='custom', pitch_length=pitch_length6, pitch_width=pitch_width6, line_color='#677ad1', **pitch_spec) TITLE_HEIGHT = 0.1 # title axes are 10% of the figure height # width of pitch axes as percent of the figure width TOP_WIDTH = 0.27 BOTTOM_WIDTH = 0.18
import matplotlib.pyplot as plt # read data df = read_event(f'{EVENT_SLUG}/7478.json', related_event_df=False, shot_freeze_frame_df=False, tactics_lineup_df=False)['event'] ############################################################################## # Filter passes by Jodie Taylor df = df[(df.player_name == 'Jodie Taylor') & (df.type_name == 'Pass')].copy() ############################################################################## # Plotting pitch = Pitch() fig, ax = pitch.draw(figsize=(8, 6)) hull = pitch.convexhull(df.x, df.y) poly = pitch.polygon(hull, ax=ax, edgecolor='cornflowerblue', facecolor='cornflowerblue', alpha=0.3) scatter = pitch.scatter(df.x, df.y, ax=ax, edgecolor='black', facecolor='cornflowerblue') plt.show( ) # if you are not using a Jupyter notebook this is necessary to show the plot
'connectionstyle': 'angle3,angleA=0,angleB=-90', 'color': FONTCOLOR } font_kwargs = { 'fontsize': 14, 'ha': 'center', 'va': 'bottom', 'fontweight': 'bold', 'fontstyle': 'italic', 'c': FONTCOLOR } for idx, pt in enumerate(pitch_types): if pt in ['tracab', 'metricasports', 'custom', 'skillcorner']: pitch = Pitch(pitch_type=pt, pitch_length=105, pitch_width=68, **pitch_kwargs) else: pitch = Pitch(pitch_type=pt, **pitch_kwargs) pitch.draw(axes[idx]) xmin, xmax, ymin, ymax = pitch.extent if pitch.dim.aspect != 1: TEXT = 'data coordinates \n are square (1:1) \n scale up to a real-pitch size' axes[idx].annotate(TEXT, xy=(xmin, ymin), xytext=(0 + (xmax - xmin) / 2, ymin), **font_kwargs) axes[idx].xaxis.set_ticks([xmin, xmax]) axes[idx].yaxis.set_ticks([ymin, ymax]) axes[idx].tick_params(labelsize=15) if pt == 'skillcorner':
# I created a function to calculate the maximum dimensions you can get away with while # having a set figure size. Let's use this to create the largest pitch possible # with a 16:9 figure aspect ratio. FIGWIDTH = 16 FIGHEIGHT = 9 NROWS = 1 NCOLS = 1 # here we want the maximum side in proportion to the figure dimensions # (height in this case) to take up all of the image MAX_GRID = 1 # pitch with minimal padding (2 each side) pitch = Pitch(pad_top=2, pad_bottom=2, pad_left=2, pad_right=2, pitch_color='#22312b') # calculate the maximum grid_height/ width GRID_WIDTH, GRID_HEIGHT = pitch.calculate_grid_dimensions(figwidth=FIGWIDTH, figheight=FIGHEIGHT, nrows=NROWS, ncols=NCOLS, max_grid=MAX_GRID, space=0) # plot using the mplsoccer grid function fig, ax = pitch.grid(figheight=FIGHEIGHT, grid_width=GRID_WIDTH, grid_height=GRID_HEIGHT,
# Filter dataset to only include one teams passes and get boolean mask for the completed passes df_pass = df.loc[mask_team1, ['x', 'y', 'end_x', 'end_y', 'outcome_name']] mask_complete = df_pass.outcome_name.isnull() ############################################################################## # View the pass dataframe. df_pass.head() ############################################################################## # Plotting # Setup the pitch pitch = Pitch(pitch_type='statsbomb', pitch_color='#22312b', line_color='#c7d5cc') fig, ax = pitch.draw(figsize=(16, 11), constrained_layout=True, tight_layout=False) fig.set_facecolor('#22312b') # Plot the completed passes pitch.arrows(df_pass[mask_complete].x, df_pass[mask_complete].y, df_pass[mask_complete].end_x, df_pass[mask_complete].end_y, width=2, headwidth=10, headlength=10, color='#ad993c',