**Data Visualization Analysis of English Premier League Team Performance**
\*\*This is a Double-blind review, please do not include authors information in this version \*\*
Received Date: \*date
Accepted Date: \*date
Published Date: \*date
**HIGHLIGHTS**
- Data Visualization analysis for English Premier League (EPL) from season 2014 - 2015 until 2021 - 2022
- Data analysis of team and players performance including team standings and results, team performance statistics, and player rating statistics
- The requirements must provide good visualization in terms informative, enjoyable to use, and usable
- Developed EPL Team and Players dashboard using Microsoft Power BI
ABSTRACT
*Football is one of the world's most popular sports, with thousands of amateur and professional players taking part every day. As a result, massive amounts of data are unavoidable. All this information is critical for many parties, including coaches, players, managers, betting companies, and the media, to analyse. However, simply viewing league standings and statistics may make it difficult for the average citizen or football fan to analyse and learn detailed information about football statistics and history. This is because most football statistic websites and sports data visual studies use basic infographics that provide limited insights into the data or use complicated methodologies that are difficult to understand for public users. In today's modern data-driven world, it's critical to identify the best, most self-explanatory way to present data, so that visual patterns that relate to the underlying available data can be formed. The goal of this project is to design and develop an English Premier League data analysis using a data visualization approach. Users' usability testing has been carried out to ensure that the project's objectives are met. The perceived usability score indicates that the website is easy to use, and the total mean of user satisfaction indicates that site visitors are impressed. According to these two general indicators, users are pleased with the website and may use it to learn about English Premier League statistics. As a result, the study successfully developed an English Premier League dashboard using a data visualization approach for analysing football data and discovering patterns, connections, and ideas.*
*Keywords: English Premier League, Sport Data Analysis Data visualization, Business Intelligent*
# INTRODUCTION
The English Premier League (EPL) is one of the world's biggest and most well-known sports leagues, with millions of supporters worldwide (Elliott, 2017). The Premier League was established on February 20, 1992, from of a vote by clubs in the Football League First Division to split from the Football League, which was founded in 1888. The Premier League, which consists of twenty clubs, is the top tier in the English football league system. Throughout the season from August to May, for a total of thirty-eight games, each side plays each other twice, once at home and once away and in total, they makes 380 matches per season (Summerfield, 2022).
Every year, massive amounts of data are generated such as overall league standing and player statistic. This information is frequently available on football websites such as the English Premier League site and Sports References site. However, it might be difficult for the average citizen or football fan to analyse and learn detailed information about football statistics and history by just seeing league standings and statistics tables without visualization. This is because, the majority of football statistic websites and sports data visual studies employ basic infographics (Theagarajan et al., 2018) that provide limited insights into the data or use complicated methodologies (Andrienko et al., 2018) that are tough for a non-visualization professional to understand.
As a result, researchers should develop interactive and easy-to-understand visualization to support the analysis and simplify information insight of English Premier League statistics. Data from football matches and the compilation of fundamental football statistics (Theagarajan et al., 2018) serve as the foundation for further analysis and visualization, such as evaluating player performance during and after a match. Because they use humans' perceptual ability to recognize visual patterns, visualizations are powerful tools for supporting the quick analysis of information (Healey et al., 2011). This might be beneficial in observing the progress of a team performance.
This paper proposes a visual and interactive dashboard based on the data of English Premier League statistics obtained from football statistics and history websites known as FBref.com and Kagge.com The dashboard shows correlations between overall data statistics such as overall league standing result, comparison of team and player performance, and expected performance. This paper uses Microsoft Power BI to create the interactive data visualization dashboard. Microsoft Power BI is a type of software program and tools that work together to transform data into interactive visualizations with an easy-to-use interface that allows end-users to build their reports and dashboards (Ferrari & Russo, 2016).
# RELATED WORK
Perin et al. (2018) review current accomplishments and challenges in sports data visualization. This sports data visualization study covers a wide range of visualization activities and objectives. Designing new visualization techniques, adapting current visualizations to a new area, and performing design studies and assessments in close collaboration with professionals such as practitioners, enthusiasts, and journalists. This study examines existing research contributions via the perspective of three types of sports data: box score data (data comprising statistical summaries of a sporting event such as a game), tracking data (data describing in-game activities and trajectories), and meta-data (data about meta-data) (data about the sport and its participants but not necessarily a given game).
Malqui et al. (2019) develop a visual analytics system to analyse the sequence and trajectory of consecutive passing sequences. The system uses a two-phase clustering algorithm that extracts typical trajectory clusters in passing sequences, which result in eight predominant clusters. The combined analysis of the sequence and trajectory clusters allows experts to perform multi or single-game analysis in multiple methods. The researcher shows the potential of their approach in case studies using data from the Brazilian and Turkish leagues and reports feedback from soccer experts.
Theagarajan (2018) propose a method for automatically generating visual analytics from soccer footage to assist coaches and recruiters in identifying the most potential players. To generate visual analytics, Convolutional Neural Networks (CNNs) is used to localize soccer players in a video and identify ball-controlling players, Deep Convolutional Generative Adversarial Networks (DCGAN) for data enhancement, a histogram-based matching to identify teams, and frame-by-frame prediction and verification analyses. The researchers then compare their method against state-of-the-art methods, achieving an accuracy of 86.59 percent in detecting ball-controlling players and 84.73 percent in generating game analytics and player statistics. This study was limited to the general audience and provided little insight into the data. The method is too complicated, and the visualization is too plain for non-visualization experts to comprehend.
Burch et al. (2020) propose a visual and interactive dashboard to aid in the extraction of insights from soccer data in this study. The dashboard depicts relationships between overall data, such as goals scored or received, ball possession, and final position after the tournaments. As a data source, the researcher used several databases including information from all FIFA World Cups. Results of every World Cup soccer match, including the half-time score for each game, referees, and stadium. Statistics for each World Cup, include the number of countries that advanced to the semi-finals, the total number of matches and goals scored, and the number of people that attended the competition.
The researcher focuses on providing an overview of the information in the dashboard, which includes all teams and World Cups, before narrowing it down to top-performing countries and looking at specific national team information. In a variety of ways, teams are compared. Our goal is to be able to identify the most successful teams and analyse their capabilities based on a variety of factors. Further restricting the scope, the researcher hopes to use the data to examine the teams' goal-scoring habits and identify their strengths and flaws. At this stage, individual players play a role since we know who scored goals and where they came from. This opens the possibility of deducing team strategy knowledge from goal-related data.
Many earlier related works, on the other hand, employ either complex methodologies that are difficult for a non-visualization person to understand, or they only use simple infographics that provide limited information from data. Many visualizations and fundamental component views are usually overlooked in these dashboards. Furthermore, most football dashboards and visualizations employ the same fundamental data charts. Static measurements are frequently displayed as a simple graphic or a comparison spreadsheet. While there are new and innovative visualizations. The bulk of existing ones concentrate on a few aspects of the data and so do not provide various perspectives on it. Filtering certain matches or players, scrolling through an attack timeline, or viewing a depiction of the shots on target are all actions that may aid in match analysis.
In conclusion, based on the literature research, a variety of suitable techniques and technologies can be implemented for this project. For example, a proper data visualization approach and interactive dashboard may aid in seeing overall statistics and performance more effectively and appealingly, allowing viewers to compare their team's chances of winning the game to those of other teams. Aside from that, the data visualization dashboard will assist people in learning fascinating information about a sport data analysis that they were previously unaware of. The English Premier League dashboard will let the public and football fans understand further English Premier League statistics, particularly data connected to their favorite team.
# METHODOLOGY
## Requirement Analysis
Requirement analysis is conducted through literature review. Data visualization effectiveness is depending on five data properties: desirable, usable, effective, measurable, and scalable. Data visualization of high quality comes in a variety of sizes and styles, but each has its own set of characteristics that ensure it provides accurate data insights. In general, the result of a good visualization must be informative, in the sense that the user will use it regularly and be able to make an effective choice based on a situation; desirable, in the sense that it must be enjoyable to use; usable, in the sense that the user will be able to meet their goal simple and fast. The visualization program must perform in the same manner as the data amount grows. As a result, the system must be scalable and measurable to accommodate future changes.
There are five main pages in the dashboard:
- Main Page. Home page and entry point to other pages.
- Matches Result & Fixture. Provide matches and teams information such as team name, match time, and the venue will be displayed.
- League Tables & Statistic. Provide a list of EPL teams, and their rankings based on their performance in each season and related information such as points, wins, losses, and goals will be displayed.
- Team Performance. Provide team's performance in the Premier League based on performance metrics such as squad standard statistics, goalkeeping, shooting, passing, and possession.
- Team Analysis. Provide squad line-up and formation, allowing the user to analyse team performance so that it can be used to decide in team tactical options based on the player performance.
## Data Visualization Design
The study's third step is the design phase. The first task in this phase is to create a dashboard mock-up and storyboard. The dashboard will show a graphical representation of data from the EPL. The mock-up and storyboard will be sketched at this point to show how the user interface design process works. Figma software was used to create the storyboard and mock-up. This is to illustrate the design before the creation of the final dashboard and to identify relevant visual content.
## Data Collection
The developer used two open data sources: EPL’s dataset from FBref.com and FIFA 22 complete player dataset from the Kaggle.com website. The dataset includes EPL data such as the standings and results, team performance statistics, and player rating statistics. These files are separated into year-by-year information. This paper chooses to get data of EPL teams from year 2015 to 2022.
The FBref.com dataset is used to create a data visualization of the English Premier League table and team performance statistics by year. Figure 6 shows the Premier Leagues Stats dataset from the FBref.com website.
Figure 1: Sample dataset from FBref.com website
Figure 7 shows the raw dataset of FIFA player rating from the Kaggle website. The dataset is in Microsoft Excel file type and contains 391 rows and 110 columns. It contains FIFA player rating dataset from 2015 to 2022 to be used to visualize player rating performance by year.
Figure 2: Dataset of FIFA player rating from 2015 to 2022
## Data Pre-Processing Stage
Pre-processing refers to any manipulation of raw data to prepare it for another type of production. It's also known as data pre-processing, which is the act of converting data into a format that makes it easier for the developer to handle. Each data attribute is transformed into a format and name for this project, making it easier for the developer to create the dashboard. Data cleansing and management will be applied to the data set collected. Based on parameters such as year, the developer will choose the appropriate and required data for display in the dashboard. The dataset will be converted to CSV format when all the data cleaning is completed, making it easy to import into Microsoft Power BI**.** Figure 8 shows the dataset of the English Premier League table that has been processed.
Figure 3**:** Dataset of English League table
## Data Transformation
This work uses Microsoft Excel and Microsoft Power BI to transform one sheet of data into distinct sheets by year and team. This is to generate visualization when the undesirable column and row remain together. Furthermore, this work able to readily view and manage data by year and team once the data was modified and transformed into multiple sheets. Developers use the filtering function in Microsoft Excel to filter unwanted data. After that, develop will select and copy the needed data into a new file based on the needed variable.
## Dashboard Development
This paper uses Microsoft Power BI for the main development tool to create data visualization and dashboard. The first step in development is to acquire data access by importing the dataset file from data sources in Microsoft Excel Workbooks file format. After importing the file, a blank workspace is created in Power BI that can be used to create data visualization.
Microsoft Power BI provides several sorts of charts, such as bar charts, column charts, stacked bars and columns, pies, half donuts, line graphs, area charts, and waterfall charts. The right pane also has a fields section that allows developers to select and switch between different data fields.
While the developer has built the dashboard and is off to a good start, it is also necessary to choose which type of visual would be most suited for which piece of data. If not thoroughly investigated, data visualization can be misleading. After the developer has developed the dashboard and filled it with useful charts, click the save button on the toolbar. Because Power BI charts are built on Excel data, they will continue to update as the Excel sheet is updated. Developers can also publish the dashboard to the Power BI service and share the link with others.
## Testing Phase
The testing phase is the fifth step. The data dashboard that has been built will be evaluated in this step. After the respondent has finished reviewing the English Premier League dashboard, a usability testing survey will be sent to them. The questionnaire's goal is to get feedback from users on the dashboard's usability and if it meets their needs. To understand how the dashboard works, users must first browse it. The URL of the questionnaire and the English Premier League dashboard will be shared on social media platforms such as WhatsApp, Telegram, and Instagram. The questionnaire is built with Google Forms to allow users to provide comments directly on the site. The questionnaire is organized into three sections: demographic background perceived usability and perceived user satisfaction. The questionnaire involves 35 respondents that asked to visit the English Premier League Dashboard before filling the questionnaires in the Google Form.
# RESULT
## Site Map
A site map is a visually or textually structured model that depicts the application's structure. The site map was an important part of a project while building a dashboard or system application. This is because the user interface hierarchy provides a visual representation of the specific data space, and its objective is to aid users or customers in knowing where they might move via the program. Furthermore, it provides visitors or clients with the convenience of quickly reviewing the full page and information. Figure 9 shows the dashboard site. It contains four main pages: League Table, Club Payer Rating and Statistic.
Figure 4 Dashboard Site Map
## User Interface Design
User interface (UI) design is an approach for building an interface for users to interact with a dashboard. The primary purpose of this UI design is to aid consumers in comprehending vital how data is presented in an entertaining and easy-to-use manner. It is critical to create a user interface that meets the expectations of users in terms of simplicity of use, user friendliness, and uniqueness. Big data refers to unusually large volumes of data that are difficult to manage using typical approaches. A dashboard with too much data may become sluggish, but a dashboard with little data can lose its utility. This section contains all of the dashboard user interface designs for the English Premier League dashboard. There are 5 main pages on the dashboard, which are homepage, English Premier League table page , English Premier League club page, and player rating detail page, and statistic page.
## Implementation
### English Premier League Table Page
The English Premier League page contains information of the English Premier League standings for the range of season. The information is presented as score card at the top of page and in a table format at the centre. The data can be sorted by any column in the table. This table contains ten items: rank, club, win, loss, draw, goals for, goal againts, points, top scorer, and season. All of this information is used to determine which team will be crowned champion at the end of the season. Additionally, using the offered filter features, the user may see the ranking of each team for the specified year. The user may identify their favourite team's performance over the course of the season. Figure 10 illustrates the table page for the English Premier League.
Figure 5: English Premier League table page
### Club Performance Data Visualization
The Club Statistic page contains overall team statistic card, big chances created, goal per match, shots, and shot on target. The team score card contains total of number of big chances created, goal per match, shots, shot on target, and shooting accuracy for all teams. Figure 6 depicts the user interface of club’s performance statistic data visualization.
The first data visualization presents the total number of big chances created for each team and season. The data presented using stacked bar chart. It allows to compare the performance among teams. The wider the bar the better the performance of the team compared to another teams.
The second data visualization at the right side is goal per match data visualization. The data is presented using treemap data visualization. It eases user to compare the performance among team by the number of goals created. The larger the box of the team compared to another team means the team is better in term of higher number of goals created.
The third data visualization at the left bottom of the page presents performance in term of shots created among teams from season 2014 - 2015 until 2020 – 2021. Data is presented using ribbon chart which allow users to analyse the performance by comparing the number shot among teams and the trends from season to season. It also eases users to find which team’s performance is more consistent by comparing the team’s ribbon. Team that has flatter ribbon is more consistent that other teams that has higher up and down ribbon. For example, Manchester City has more consistent performance because their ribbon is flatter than another teams’ ribbon.
Figure 6: Club’s Performance Statistic Data Visualization
The last data visualization is stacked area chart data visualization at the bottom right side of the page that present shot on target performance among teams and in each season. It allows to compare performance of team in term of number of short on target. From the data visualization, it shows that Southampton and Leicester City are top two team that have highest number short on target in every seasons.
As overall performance comparison from this page, it shows that Manchester City is the best team in the performance criteria of big chances created, goal per match, and shots. It is followed by Liverpool, Chelsea and Tottenham Hotspur.
### Player Rating Detail Page
The player rating page contains detailed information on each player, including player’s biodata and performance details. The page allow user to select which player to be displayed by player’s short name, club, club position, and nationality. Then, the player’s biodata data such as weight, height, jersey number, and age are appeared at the right side. The players performance data are presented at the right bottom of the page. The user may use to find and compare prior and current season player rating performance. This paper presents player’s data visualization to present each aspect's measurement using gauge to determine which rating range represents good (green), medium (yellow), and poor (red) performance for each player. Figure 11 depicts the user interface for the player rating page.
Figure 7: Player rating detail page
# FINDINGS AND DISCUSSIONS
In the finding and discussion part, a questionnaire technique was employed to get relevant input from a varied sample of consumers. The study participants were football lovers and non-visualization expert adults who visited the English Premier League dashboard. Surveys are divided into two categories: usability and user satisfaction. Figure 8 presents the questionnaire result of users’ perceived usability evaluation and Figure 9 presents the questionnaire result of users’ satisfaction evaluation.
\[CHART\]
Figure 8: Perceived Usability Evaluation Result Graph
\[CHART\]
Figure 9: Perceived User Satisfaction Evaluation Result Graph
The total mean for the users’ perceived usability is 4.45 and users’ satisfaction is 4.45. These indicating that users are satisfied with the website usability. According to these two general indicators, users are pleased with the website and may utilize it for analysis of English Premier League.
**CONCLUSION AND RECOMMENDATIONS**
Several restrictions were encountered while building the project. The developer attempted to obtain all the datasets on the player and club from the official English Premier League website but was unable to do so owing to a limited data set. The data set from the official English Premier League website was manually collected and entered into Microsoft Excel. This project took data from an offline dataset that was not constantly updated. The graphical depiction of the English Premier League dashboard only displays the datasets that are currently available. As a result, the dashboard system for the English Premier League was unable to update the real-time data.
Along with the constraints indicated, there are various recommendations for further research on the English Premier League data visualization dashboard. Future implementation of this recommendation could make the project more effective and efficient. The project might be enhanced by using a different approach to extract the data. Manual extraction yields a limited quantity of data. Furthermore, leveraging real-time datasets and allowing live changes to visual graphics can improve the project. Improving the dashboard with more interactive features and information may also benefit the project. Dashboards can be linked with or provided on website platforms to enable additional interesting functions. A user's account can be set up to allow them to manage their filtered info which includes materials that encourages users to connect and communicate.
In conclusion, this study has met all its objectives, including establishing the requirements and strategies for developing a data visualization dashboard by creating and designing the English Premier League dashboard in Microsoft Power BI. The data was successfully presented in a dashboard and published online via the Microsoft Power BI link to be shared on the social media platform, allowing football fans and non-visualization experts to analyse and understand English Premier League statistics. Finally, future iterations of the project may be enhanced and made more functional by implementing this recommendation.
# REFERENCES
Andrienko, G., Andrienko, N., Budziak, G., Von Landesberger, T., & Weber, H. (2018). Exploring pressure in football. *Proceedings of the Workshop on Advanced Visual Interfaces AVI*, 1–3. https://doi.org/10.1145/3206505.3206558Burch, M., Wallner, G., Angelescu, S. L., & Lakatos, P. (2020). Visual Analysis of FIFA World Cup Data. *2020 24th International Conference Information Visualisation (IV)*, 114–119. https://doi.org/10.1109/IV51561.2020.00028Elliott, R. (2017). *The English Premier League: A socio-cultural analysis*. https://doi.org/10.4324/9781315636696Ferrari, A., & Russo, M. (2016). *Introducing Microsoft Power BI*. Microsoft Press.Healey, C. G., Member, S., & Enns, J. T. (2011). *Attention and Visual Memory in Visualization and Computer Graphics*. 1–20.Malqui, J. L. S., Romero, N. M. L., Garcia, R., Alemdar, H., & Comba, J. L. D. (2019). How do soccer teams coordinate consecutive passes? A visual analytics system for analysing the complexity of passing sequences using soccer flow motifs. *Computers & Graphics*, *84*, 122–133. https://doi.org/10.1016/j.cag.2019.08.010Perin, C., Vuillemot, R., Stasko, J. T., Wood, J., & Carpendale, S. (2018). *State of the Art of Sports Data Visualization—Perin—2018—Computer Graphics Forum—Wiley Online Library*. https://onlinelibrary.wiley.com/doi/abs/10.1111/cgf.13447Summerfield, J. (2022, June 20). *How many games are in a Premier League season?* 90min.Com. https://www.90min.com/posts/how-many-games-are-in-a-premier-league-seasonTheagarajan, R., Pala, F., Zhang, X., & Bhanu, B. (2018). Soccer: Who Has the Ball? Generating Visual Analytics and Player Statistics. *2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)*, 1830–18308. https://doi.org/10.1109/CVPRW.2018.00227