UOU ATHLETICS - SPORTS ANALYTICS
UoU Athletics – Sports Analytics
Project Objectives and History
Our project was sponsored by the University of Utah Women’s Soccer team. The scope of our work
for the Utah Sports Athletics Project includes all planning, execution, and implementation for a
new portal capturing and analyzing athletes’ data. Below points show the lifecycle of our project:
1. Web development: The objective is to develop a user-friendly GUI using web forms so,
the data related to performance enhancement of players can be easily gathered.
2. Database implementation: The objective is to transfer the existing forms data to easily
accessible online repository as well as fetching the data from the web forms and pushing
it into database.
3. Quantitative analysis: After storing the data, the objective is to predict the future risk of
injury in order to improve the wellness of players and subsequently the performance. We
will be using predictive analytics and machine learning algorithms for the same.
The University of Utah Women’s Soccer team is a university funded sports team who competes
in the PAC 12 Conference and recently completed their 2017 season. The women’s soccer team
has been sponsoring capstone projects with the MSIS program for more than two years, allowing
this project to mature and grow. This allows us to build on an established foundation and move
forward making improvements and increasing effectiveness of the system.
The overall goal of the project itself is to track key performance indicators to use for analytics and
be able to view trends in player performance.
We had 2 strategic objectives for this project:
1. To increase the performance of the players
2. To reduce the risk of injuries of the players
Our Team
The team we have for this project consists of four people who come from various backgrounds
and levels of experience. The team communicates frequently each week using email and text
messages. The team also meets with the project sponsor bi-weekly for one hour to show what
has taken place in the last week and make goals for the upcoming week. A team meeting is held
at least one time per week where all group members come together and work on the project as a
whole and address any roadblocks.
All the members have worked on different technologies our project had addressed. This allows
for everyone to understand how the project is functioning and gives each person experience with
a new technology/system. We all have worked on this project equally and have tried to work on
problem solving whenever needed, however our roles and main focuses are as follows:
Brandon Kennedy: Web Development
Gaurav Kutemate: Data Visualization and Predictive Analytics, Incorporating Polar API
Kanika Moondra: Data Visualization and Predictive Analytics, Incorporating Polar API
Sakshi Vig: Database Integration, Web Development
Project Approach
As a team we decided to employ some aspects of Agile development including regular standups,
weekly sprints, and retrospectives. This enabled our team to get working quickly and facilitated
the flow of communication between team members. Using this structure also helped us in
overcoming roadblocks efficiently and allowing us to continue working on the goals of the sprint.
Using this method helped us break down large goals into reasonable tasks and provided a timeline
for all team members and the sponsor.
When we took on this project, the system was completely broken, and the team wasn’t able to
use the web forms to enter data. We first tried to look at the code that past teams have used but
we decided to build everything from scratch as it would have taken us an equal amount of time
to modify their code as to write a new one. Also, the past groups had just one production
environment against a good development practice i.e. create Test and Product environments
before deployment. So, we decided to have 2 environments: test and production so we can
always resume of previous working version.
We created new web forms so that it becomes easy for players, trainers and coaches to fill their
data.
We used MySQL workbench to integrate current data with legacy data so that value of legacy
data exists.
Below are the technologies that we used in our project:
● Data capture/web form
○ .Net Core
○ Vue.js
● Data storage
○ MySQL Workbench
● Data analysis and Visualization
○ R
○ Power BI
● Communication and tracking
○ Google products
○ Text Messages
Overall, the team is very happy about the experience we had with the project and the tools we
were able to learn and put to use. We were excited to face challenges of starting the project from
scratch and finding solutions to fix and improve all the challenges from the past. This provided a
very “real world” experience for us and has been immensely helpful in our current places of
occupation. Now we will discuss in more detail the results and lessons learned from each portion
of our project.
Data Capture/Web Form
The web application is the main entry point for data in the application. There are forms for
coaches, trainers and athletes to enter information about the athletes’ wellness and performance.
The web application has had several iterations, each building upon the last. It started out as Excel
spreadsheet then migrated to Google forms and finally became a stand-alone web application.
When we took over the application it was not being used. The application was not working, and
the sponsors had switched back to using the google forms to gather data.
After speaking with the sponsor about the application and doing some of our own analysis, we
determined that the primary objective should be to make the application more reliable, as the
availability of the web application had caused challenges in driving the adoption of the application.
We determined that improvements could be made to the application, the infrastructure as well as
the development process that would make the application more reliable.
Objectives
To improve the application we decided to work on the following objectives:
1. Unify the source code. The application was made up of two different code bases. It was
originally written in ASP.net and later a group started writing some of the code in Java,
with the intention of future groups converting entire application. This added complexity to
the development process as well as administration of the application. Whenever the
server was restarted for updates or power outages, part of the application had to be
manually restarted.
2. Add all code to source control. Not all of the code for the application was in source control.
This was one of the main reasons that the code was not working. There was no way to
revert to a previous version of working code.
3. Create multiple development environments. In the previous state of the application there
was only one environment, the production environment. There was no way to test
changes to the code base without pushing the code to production. A second test
environment would allow the team to thoroughly test the application before deploying it to
the production environment.
4. Update outdated and deprecated components. Many of the components used for the
application were outdated and some were deprecated. Of note the MySql client used to
access the database was outdated and could not be used to connect to a newer version
of MySql server. Because of that the database server could not be upgraded.
5. Improve the user interface. The user interface needed some improvements to make it
easier to input large amounts of data. One of the biggest problems was that some forms
had to be submitted for each athlete that was being tracked each day.
Rewriting the Application
We spent some time determining how to achieve our objectives and we determined that the best
option would be to rewrite the application. Rewriting the application was a drastic option, but we
felt that it made sense. First off we didn’t need to start completely from scratch, the major design,
the lessons learned and even some of the code could be leveraged to create the new application.
Also, updating the components of the application was going to take a lot of effort. There were
some components that were deeply embedded in the application and would take substantial effort
to replace with newer versions.
There were other benefits of rewriting the application as well. Firstly, rewriting the application
would allow us to create a mobile-first application. This kind of application would provide a lot of
benefits to student athletes who would be fill out their daily wellness forms from their mobile
phones instead of a computer. Finally, newer application frameworks are better at tracking and
keeping code libraries updated. Utilizing these capabilities could help keep outdated components
from causing issues from again in the future.
Changes to the Trainer Form
One of the major pain points of the previous application was the trainer form. The form was time
consuming to fill out for a couple of reasons. First of all, despite the fact that some fields of data
in the form that were not meant to be collected every day, they were still included in the form.
Each of those fields had to be filled out every day so they had to be left blank which left holes in
the data. The other problem with the trainer form was that it had to be filled out for each athlete
who participated in the training, information like what date the training occurred had to be selected
for each athlete.
Previous Trainer Form
To improve the form, we made a couple of changes. First of all, we broke up a single form into
three different forms, Injury, USG rating and Vertical jump. The change allows the trainer to fill
out only the forms that are necessary based on which training was done that day. The vertical
jump for example is not measured every day, so there is no reason to have to input it every day.
Next we changed to the form to allow the trainer to enter the information for all of the athletes on
the team at one time. This removes the need for the trainer to enter the form individually for each
athlete so that they don’t have to enter duplicated information for every athlete.
USG Form
Vertical Jump Form
Player Injury Form
Data Integration
As this project is ongoing for past 5 years, there were different platforms used to capture the data.
Various platforms were:
Excel → Google forms → Web Forms → Google Forms
The data was initially captured in excel sheets and then, moved to google forms in order to
integrate that with google studio for analysis. Furthermore, in 2017 a platform was designed to
capture the data using web forms by designing a complete web application, hence on backend
MySql was used to capture the data. As the system stopped working, the team had to go back to
prior functioning stage of google forms.
We tried to create a homogenous & single data source for analysis, but it was big challenge to
populate the data accurately. Halfway through the project we realized the attributes to capture
data have changed for better i.e in order to do the detailed analysis the design of forms to capture
data and fields have changed so it is not feasible to integrate into a single database. In order to
understand the data better and analyze, we created different schemas to store all the previous
data.
This not only helped us keep the previous data but also provide sufficient sample size to do
detailed analysis.
Now the new system is up and running, we have used MySql at backend and databases are as
follows:
Current New Database:
Google Form Database :
Data Analysis and Visualization
We used Power BI and R for analyzing data and digging out meaningful insights. When we began
looking at what the previous group had done, we saw that they gave us a decent framework to
start with. However, we also realized that there were many things we needed to accomplish before
we could get to the point of moving forward with our stated objectives.
We pulled the google forms data that has been incorporated to database in order to create
visualizations.
Here are some of the visualizations we created to meet our 2 strategic objectives:
Improving performance of the players
Fitness Details:
The below chart shows the average of different fitness attributes of players by month. From this,
we observed the peak is in May which makes quite a lot of sense because that’s the start of the
Season and players are under intense training during that period.
This graph shows the average Wellness Scores for players which is actually a summarization of
the above graph attributes and we can see from this graph as well reflects the peak during the
month of May.
Correlation Plot:
Below is the correlation plot between different player attributes in order to understand which
attributes are highly related to each other and what is their strength of correlation. For example:
We can see that Muscle soreness and Nutrition are highly correlated with a strength of 0.9.
In a similar manner we can see how other attributes are related to each other. To statistically
prove this, we created a linear regression model keeping Muscle Soreness as our target
variable as we feel that player’s performance is directly related to how her muscle strength is
and how fast their muscles get recovered after the game.
Linear Regression Model:
The model shows as nutrition increases by 1 unit, muscle soreness decreases by 0.04 which is
an indicator of player performance. The conclusion that can be driven from this is that the better
players eat, the better they will be able to perform on the field.
Reducing the risk of injuries
In order to reduce the risk of injuries, it is really important to understand 3 factors:
- What kinds of injuries occur and the count of injuries?
- Where does the injury occur?
- What are the causes of injury?
Below chart addresses the first point showing the count of injuries by month and their description
Next graph shows where does the injury occur and we can see that maximum percent of injuries
occur due to Soccer Training which is quite obvious as soccer training involves a much intense
workout session and many muscles are involved at the same time thereby increasing the risk of
injury.
We tried to dig deeper into Soccer Training to understand what the causes of injuries are and
found that non-contact injuries account to max. Example: Ligament tear, Muscle sprain, fractures
by running, jumping etc.
If the players can prevent these injuries, they will be able to reduce their risk of injuries on the
field.
Data Flow
Below is the gist of our project and how all components are connected to each other. Players,
trainers and coaches will input data via in-house web forms which will then be pulled into database
and further be pulled into Power BI for creating insights.
Data input Data Storage Data Analytics
Future Scope:
1. Integration of Polar device API: We tried to incorporate Polar API with our current
system but faced few challenges which can be addressed in future. It will help the team
get their physical data like Heartbeat, Blood pressure, Sleep time etc.
2. In-Stat Reports: The team has signed for ‘instat’ reports which means the whole game
will be recorded by camera which can later be used to understand what affected the final
result of the match, playing style of opponent, calcuating number of opponent attacks,
their direction, effectiveness, etc.
With these multiple data sources and our system in place, data collection and analysis can
become quite comprehensive and reliable.