Soccer Through the Ages
This dataset contains information on international soccer games throughout the years. It includes results of soccer games and information about the players who scored the goals. The dataset contains data from 1872 up to 2023.
💾 The data
data/results.csv
- CSV with results of soccer games between 1872 and 2023home_score
- The score of the home team, excluding penalty shootoutsaway_score
- The score of the away team, excluding penalty shootoutstournament
- The name of the tournamentcity
- The name of the city where the game was playedcountry
- The name of the country where the game was playedneutral
- Whether the game was played at a neutral venue or not
data/shootouts.csv
- CSV with results of penalty shootouts in the soccer gameswinner
- The team that won the penalty shootout
data/goalscorers.csv
- CSV with information on goal scorers of some of the soccer games in the results CSVteam
- The team that scored the goalscorer
- The player who scored the goalminute
- The minute in the game when the goal was scoredown_goal
- Whether it was an own goal or notpenalty
- Whether the goal was scored as a penalty or not
The following columns can be found in all datasets:
date
- The date of the soccer gamehome_team
- The team that played at homeaway_team
- The team that played away
These shared columns fully identify the game that was played and can be used to join data between the different CSV files.
Source: GitHub
📊 Some guiding questions and visualization to help you explore this data:
- Which are the 15 countries that have won the most games since 1960? Show them in a horizontal bar plot.
- How many goals are scored in total in each minute of the game? Show this in a bar plot, with the minutes on the x-axis. If you're up for the challenge, you could even create an animated Plotly plot that shows how the distribution has changed over the years.
- Which 10 players have scored the most hat-tricks?
- What is the proportion of games won by each team at home and away? What is the difference between the proportions?
- How many games have been won by the home team? And by the away team?
💼 Develop a case study for your portfolio
After exploring the data, you can create a comprehensive case study using this dataset. We have provided an example objective below, but feel free to come up with your own - the world is your oyster!
Example objective: The UEFA Euro 2024 tournament is approaching. Utilize the historical data to construct a predictive model that forecasts potential outcomes of the tournament based on the team draws. Since the draws are not known yet, you should be able to configure them as variables in your notebook.
# Imported all necessary frameworks
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns
import numpy as np
import datetime
import plotly.express as px
# load all necessary dataset
goalscorers = pd.read_csv('data/goalscorers.csv')
results = pd.read_csv('data/results.csv')
shootouts = pd.read_csv('data/shootouts.csv')
results = results.merge(shootouts, how="left")
# Created column with winners
results['win'] = None
results['win'] = np.where((results['home_score'] > results['away_score']) &
(results['win'] != None) ,
results['home_team'], results['win'])
results['win'] = np.where((results['home_score'] < results['away_score']) &
(results['win'] != None),
results['away_team'], results['win'])
results['win'] = np.where((results['home_score'] == results['away_score']) &
(results['win'] != None),
pd.Series(["draw"]), results['win'])
results['win'] = np.where((results['winner'].isin(results['home_team'])) | (results['winner'].isin(results['away_team'])) & (results['win'] == 'draw'), results['winner'], results['win'])
# Converted date with string to datetime
results["date"] = results['date'].apply(lambda x : datetime.datetime.strptime(x, '%Y-%m-%d'))
1.The 15 countries that have won the most games since 1960
# Created dataset with winners since 1960
results_since_1960 = results[results["date"] >= datetime.datetime.strptime('1960-01-01', '%Y-%m-%d')]
won_since_1960 = results_since_1960['win'].value_counts().drop(index='draw').head(15)
won_since_1960 = pd.DataFrame(won_since_1960).reset_index(names='country')
sns.set_theme(style="whitegrid")
sns.barplot(won_since_1960, x='win', y='country').set(title='The 15 countries that have won the most games since 1960')
# Converted date with string to datetime
goalscorers['date'] = goalscorers['date'].apply(lambda x : datetime.datetime.strptime(x, '%Y-%m-%d'))
goalscorers['year'] = goalscorers['date'].dt.strftime('%Y')
# Created df with minutes and goals
minute_goals = goalscorers[['minute']].groupby(['minute']).size()
# Built barplot
plt.figure(figsize=(20, 6))
plt.xticks(rotation=90)
sns.barplot(goalscorers, x=minute_goals.index, y=minute_goals.values)
plt.xlabel('Minute')
plt.ylabel('Total Goals')
plt.title('Total Goals Scored in Each Minute of the Game')
plt.show()
# Group the data by year and minute and calculate the total goals
minute_goals = goalscorers[['year', 'minute']].groupby(['year','minute']).size().reset_index(name='total_goals')
# Create an animated bar plot
fig = px.bar(minute_goals, x='minute', y='total_goals', animation_frame='year', range_x=[0, 122], range_y=[0, 100],
labels={'minute': 'Minute', 'total_goals': 'Total Goals'},
title='Distribution of Goals Over the Years')
# Show the plot
fig.show()
# The number of goals scored by each football player is calculated
goals = goalscorers[['date', 'scorer']].groupby(['date', 'scorer']).size().reset_index(name='goals')
# The number of hat trick by each football player
hat_trick_count = goals[goals['goals'] >= 3][['scorer', 'goals']].groupby('scorer').size().reset_index(name='hat_trick')
hat_trick_count = hat_trick_count.sort_values('hat_trick', ascending=False).head(10)
# 10 players with the most hat-tricks per country
plt.figure(figsize=(12, 6))
sns.barplot(data=hat_trick_count, x='hat_trick', y='scorer')
plt.xlabel('Number of Hat-tricks')
plt.ylabel('Player')
plt.title('Top 10 Players with the Most Hat-tricks')
plt.show()
# Home wins counted per each country
home_wins_per_country = results[results['win'] == results['home_team']]['home_team'].value_counts() / results['home_team'].value_counts()
# Away wins counted per each country
away_wins_per_country = results[results['win'] == results['away_team']]['away_team'].value_counts() / results['away_team'].value_counts()
# Countries that have never won at home or away are set to 0
home_wins_per_country = home_wins_per_country.fillna(0)
away_wins_per_country = away_wins_per_country.fillna(0)
# Difference between home wins and away wins per each country
difference = home_wins_per_country - away_wins_per_country
‌
‌