Visualisation of Defensive Pressure in Football

SPE5AMS – Portfolio Assessment

1. Introduction

In this tutorial, we will create visualisations of Defensive Pressure in the UEFA EURO 2024 Quarter-Final match between Spain and Germany.

We’ll be using density plots to visualise the areas on the pitch where defensive pressure from both teams was most concentrated. Instead of providing an in-depth analysis of the defensive pressure characteristics between Spain and Germany, this guide will focus on how to visually present spatiotemporal data for tactical analysis. This step-by-step tutorial will help you gain insights into how to incorporate data visualisation for football tactics.”

For this tutorial, we will be using a open dataset from StatsBomb.
Before we begin, please read the User Agreement.
Also, you can check the step-by-step tutorial guide available from StatsBomb on how to use their R package.

2. Import Standard Events Data

First, install and load the required packages.

# Instal package
install.packages("devtools") 
devtools::install_github("statsbomb/StatsBombR") # Install StatsBombR package


# Import packages
library(tidyverse)
library(StatsBombR)
library(ggsoccer)
library(viridis)

You can use FreeCompetitions() function to retrieve a list of all the competitions available in the free StatsBomb datasets.

FreeCompetitions()

Step-1
From a list of all competitions, we will filter the data to select the 2024 UEFA European Championship.

Step-2
Use c function to retrieve all matches from the 2024 UEFA EURO.

Step-3
Use free_allevents() function to download the Standard Events data for all matches in the UEFA EURO.

Step-4
Now that we have the raw event data for the 2024 UEFA EURO, working with JSON files in R can be a bit challenging. To simplify this process, we can use the allclean() function, which cleans the data and adds extra useful information.

Step-5
Finally, we will filter the data to retrieve only the Quarter-Final match between Spain and Germany.

# Select the 2024 UEFA Euro 
Comp <- FreeCompetitions() %>% 
  filter(competition_id == 55 & season_id == 282)

# Check all the matches within the 2024 UEFA Euro
UFFA <- FreeMatches(Competitions = Comp)

# Download the UFFA Euro data for all the matches
EURO <- free_allevents(MatchesDF = UFFA, Parallel = T)

# Clean up the data format that is appropriate for R 
EURO <- allclean(EURO)

# Filter to get Spain vs. Germany in Quarter-Final
Spain <- EURO %>% 
  filter(match_id == 3942226)

3.1 Import 360-Events Data

Note - This is an optional part and it will walk you through how to obtain the location data of other players in each event.

In this section, we will download the 360-Events data for the UEFA EURO 2024 Quarter-Final match between Spain and Germany. The 360-Events data includes the location of all players on the pitch. This data can be used for more in-depth analyses, such as assessing pitch control.

# Download UFFA Euro 360-event data for all the matches
EURO_360 <- free_allevents_360(MatchesDF = UFFA, Parallel = T)

# Filter to get Spain vs. Germany in Quarter-Final
Spain_360 <- EURO_360 %>% 
  filter(match_id == 3942226)

# Rename the event_ID in 360 data
Spain_360 = Spain_360 %>% 
  rename(id = event_uuid)

3.2 Combine Standard Events & 360-Events datasets

Note - This is an optional part and it will walk you through how to obtain the location data of other players in each event.

We will combine the Standard Events data and 360-Events data. By merging these datasets, we can get access to the events data with information about the players’ positions on the pitch.

# Combine 360-event data & Standard event data
Player_location <- Spain %>% 
  left_join(Spain_360, by = c("id" = "id"))

# Clean the data
Player_location <- Player_location %>% 
  rename(match_id = match_id.x) %>% 
  select(-match_id.y)

3.3 Cleaning and Processing Player Location Data

Note - This is an optional part and it will walk you through how to obtain the location data of other players in each event.

In this step, we will extract the individual player positions (x and y coordinates) for each event by unnesting the freeze-frame data.

# Remove variables that we won`t use
Player_location <- Spain %>% 
  select(id, period, minute, second, type.name, team.name, OpposingTeam, player.name, 
         location.x, location.y, freeze_frame)

# Unnest the freeze-frame and extract the x & y coordinates of player positions
Player_location <- Player_location %>% 
  unnest(freeze_frame) %>%
  mutate(ff_location.x = (map(location, 1)), ff_location.y = (map(location, 2))) %>%
  select(-location) %>%
  mutate(
    ff_location.x = as.numeric(ifelse(ff_location.x == "NULL", NA, ff_location.x)), 
    ff_location.y = as.numeric(ifelse(ff_location.y == "NULL", NA, ff_location.y)))

4. Visualise Defensive Pressure

In this step, we will visualise the Defensive Pressure events from the dataset. The Pressure event represents when a player from one team applies pressure on the ball carrier of the opposing team. We will filter the Standard Events dataset to extract only the Pressure events, which will be used to visualise where on the pitch the pressure was most intense.

# Filter only Pressure events from the Standard Events dataset
Spain_press <- Spain %>% 
  filter(type.name == "Pressure") %>% 
  select(period, minute, second, team.name, location.x, location.y)

1. Visualize the Defensive Pressure Events for Spain

# Visualise Defensive Pressure areas for Spain
ggplot() +
  # Draw football pitch by using the ggsoccer package
  annotate_pitch(
    colour = "white",               # Change the color of the pitch lines 
    fill = "#001049",               # Change the pitch background color 
    dimensions = pitch_statsbomb) + # Use the pre-defined pitch dimensions for StatsBomb data
  theme_pitch() +                   # Removing the unnecessary ggplot plot details
  
  # Create the 2D density plot
  stat_density_2d(data = Spain_press %>% 
                    # Filter to retrieve Pressure event data for only Spain
                    filter(team.name == "Spain"),
                  # Map the location of a player who applied defensive pressure to opponent player
                  mapping = aes(x = location.x,      
                                y = location.y, 
                                fill = ..density..), # Fill the density color
                  geom = "raster",  # Use a raster (heatmap) style
                  contour = FALSE,  # Disable contour lines
                  alpha = 0.7) +    # Set the transparency for density color
  
  # Adjust the plot dimensions (Clip the data plotted outside of readjusted dimensions)
  coord_cartesian(xlim = c(6, 114), ylim = c(5, 75), clip = "on") + 
  # Use "viridis" color palette to fill the density plot
  scale_fill_viridis() +
  # Add titles and caption
  labs(title ="Denfensive Pressure Area for Spain",
       subtitle = "Direction of Play →",
       caption = "Data Source: StatsBomb") +
  # Adjust theme settings
  theme(legend.position = "none",  # Hide the Legend
          plot.subtitle = element_text(size = 10,         # Change the font size of subtitle
                                       color = "dimgray", # Change the font color
                                       hjust = 0.5))      # Change the position of subtitle

2. Visualize the Defensive Pressure Events for Germany

# Visualise Defensive Pressure areas for Germany
ggplot() +
  annotate_pitch(colour = "white",
                 fill = "#001049",
                 dimensions = pitch_statsbomb) +
  theme_pitch() +
  stat_density_2d(data = Spain_press %>% 
                    filter(team.name == "Germany"), # Filter to retrieve Pressure event data for only Germany
                  mapping = aes(x = location.x, y = location.y, fill = ..density..), 
                  geom = "raster", contour = FALSE, alpha = 0.7) + 
  coord_cartesian(xlim = c(6, 114), ylim = c(5, 75), clip = "on") +
  # Use "viridis (magma)" color palette to fill the density plot
  scale_fill_viridis(option = "magma") +
  labs(title = "Denfensive Pressure Area for Germany",
      subtitle = "Direction of Play →",
      caption = "Data Source: StatsBomb") + 
  theme(legend.position = "none",
        plot.subtitle = element_text(size = 10, color = "dimgray", hjust = 0.5))