add match momentum example

examples/tutorials/plot_match_momentum.py

@@ -0,0 +1,106 @@
+"""
+===============
+Match Momentum
+===============
+
+This example shows how to plot match momentum using the calculations of expected threat (xT) model shown in the same folder.
+It can be applied to any match where we have the xT values of every event that happend, the team that made it and the minute when it happened.
+
+"""
+
+import matplotlib.pyplot as plt
+import numpy as np
+import pandas as pd
+from scipy.interpolate import make_interp_spline
+
+##############################################################################
+# Import the data
+# ---------------
+# For this, we're using data from the FA Women's Super League 2019/20 as used in the `Improvements to expected threat <https://mplsoccer.readthedocs.io/en/latest/gallery/tutorials/plot_xt_improvements.html>`
+# but just limiting to one match with the xT already calculated (dataframe move_success).
+# The only thing that changes is the fact that we need the columns of "minute" and "team_name" to plot match momentum correctly.
+# You may add them where the list "cols" is defined in the other tutorial.
+
+match_events = pd.read_csv("https://raw.githubusercontent.com/andrewRowlinson/mplsoccer-assets/main/xt_match_example.csv")
+
+##############################################################################
+# Calculating the moving averages of Expected Threat
+# --------------
+# Here we need to know the threat that every team had in each minute of the match
+# For that, we need to split the match_events dataframe in two separate ones (one for each team) and group the sum of xT values
+# by minutes.
+
+teams = match_events.team_name.unique()
+team1, team2 = teams[0], teams[1]
+
+team1_actions = match_events[match_events['team_name'] == team1].groupby('minute').sum()
+team2_actions = match_events[match_events['team_name'] == team2].groupby('minute').sum()
+
+# With the two dataframes separated, we need to calculate the mean of the moving average of the Expected Threat.
+# We are using five as a parameter to take into consideration the last five minutes of the match.
+
+team1_actions['rolling_xt'] = team1_actions.xt.rolling(5).mean()
+team2_actions['rolling_xt'] = team2_actions.xt.rolling(5).mean()
+
+# Then we need to merge them by the minute so we can calculate the difference between the two values
+# which is going to give us the values of the y-axis in the match momentum plot.
+
+match_xt = team1_actions.merge(team2_actions, left_index=True, right_index=True, how='outer', suffixes=['_t1', '_t2'])
+match_xt = match_xt[['rolling_xt_t1', 'rolling_xt_t2']].fillna(0) #so the difference can be calculated correctly.
+match_xt['total_diff'] = match_xt['rolling_xt_t1'] - match_xt['rolling_xt_t2']
+
+##############################################################################
+# Plot Match Momentum
+# --------------
+# The only things we need from the events are the minutes (x-axis) and the difference between the moving averages of xT for both teams (y-axis)
+# Also, is needed that we interpolate the values.
+
+fig, ax = plt.subplots(figsize=(16,9))
+fig.set_facecolor('white')
+
+local_color = 'lightblue'
+visit_color = 'red'
+
+x = match_xt.index
+y = match_xt.total_diff
+
+X_Y_Spline = make_interp_spline(x, y)
+X_ = np.linspace(x.min(), x.max(), 500)
+Y_ = X_Y_Spline(X_)
+
+ax.plot(X_, Y_)
+
+# Then all esthetic functions for the plot to be more readable and clear
+
+ax.axhline(0) # To separate the values
+
+# To fill with one color for the positive and negative values we use this function
+
+ax.fill_between(X_, Y_, color=local_color, where= Y_ > 0)
+ax.fill_between(X_, Y_, color=visit_color, where= Y_ < 0)
+
+ax.set_xlim(0,match_xt.index[-1]) #To compact the plot in the x-axis
+ax.spines[['right','left','top']].set_visible(False)
+
+# Considerations in x and y axis.
+ax.set_xlabel('Minutes', fontsize=20)
+ax.set_yticks([]) #You can comment this line if you want the difference of values to be shown
+ax.set_xticks(range(0,match_xt.index[-1], 10))
+
+# Lastly, if we have the data of goals in out events we can make to lists: one for the minutes and another with which team scored.
+# With that we can plot a line showing every single one of them
+
+min_goals = [ 6,  9, 16, 22, 69]
+team_goals = ['Reading WFC', 'Everton LFC', 'Reading WFC', 'Everton LFC', 'Reading WFC']
+
+
+for goal in range(len(min_goals)):
+    if team_goals[goal] == team1:
+        ax.axvline(min_goals[goal], lw=2, color=local_color, ls=':')
+    else:
+        ax.axvline(min_goals[goal], lw=2, color=visit_color, ls=':')
+
+#And a title using the data of the team_name columns
+ax.set_title(f'Match momentum {team1} vs. {team2}', fontsize=26)
+
+plt.show()