3 Lecture 1 Handouts

Module introduction and Principles of Data Visualisation

3.1 Today’s Session

Introduction to the module and its objectives
Why data visualisation?
From data to visualisation

3.2 Today’s learning objectives

Recall the module learning outcomes
Explain the use of data visualisation
Describe the principles of data visualisation

3.3 Introduction to the module

3.3.1 Module aims

Introduce you to the use of data visualisation in sport data analytics.
Improve your understanding about different data visualisation methods used in sports
Understand the basic principles of data visualisation
Learn how to tailor visualisations based on the different needs of end users

3.3.2 Module learning objectives

You will be able to evaluate the different data visualisation methods available in a professional sporting context.
You will be able to formulate the steps required to visualise data in a clear and understandable manner.
You will be able to manipulate and organise data and identify the correct data visualisation methods depending on their aim and audience.
You will be able to critique for and against the use of specific data visualisation methods depending on the message they want to share and audience they target.
You will be able to use Tableau to create interactive dashboards sharing sports performance data.

3.3.3 Content

Face-to-face session
- Lectures focus on theoretical understanding of data visualisation principles and process

- Interactive sessions in which case studies will be discussed, group tasks assigned and findings presented

- Practical demonstrations on the use of R and Tableau for data visualisation

Self-study
- A variety of readings, group work tasks and practical’s

3.3.4 Assessment

You are required to complete a:

500-word written critique. You will be given a sport data visualisation and will be asked to outline the strengths and weaknesses of this visualisation. You will use published literature to support your critique (20% module grade, week 6)
Tableau data visualisation based on a publicly available sport data set. You will work through the 7-steps of data visualisation and use R and Tableau to create your dashboard. (80% module grade, week 11)

3.3.5 Software

This module makes use of R and Tableau. It is recommend you install R and R Studio on your personal computer. R and R studio are available from: https://posit.co/download/rstudio-desktop/

To gain access to Tableau on your personal computer please go to Tableau for Students to register, download the program and obtain a free licence.

3.4 What is data visualisation?

Data visualisation is the graphical representation of information and data.
Visual elements like charts, graphs and maps provide an accessible way to see and understand trends, outliers and patterns in data.
Data visualisations enable story telling.
Creating effective data visualisations requires skill and knowledge.

3.5 What is data visualisation?

3.6 What is data visualisation?

3.7 Why visualise our data?

Approximately 70% of sense receptors are in our eyes
40% of the cerebral cortex is involved in processing visual information
The visual connection to the brain has more bandwidth than other paths
Visual perception is intimately connected to understanding

3.8 Why visualise our data?

Our brain is powerful but working memory is limited
Working memory limited to a small number of “chunks”
Visualization allows us to consolidate complex statistics so we can process more data simultaneously (seeing the forest along with the trees)
The picture is not the end goal – It’s what we do with it that is important

3.9 Let’s play a game, how many 7s can you count?

On the next slide you will see a square of numbers You have 15s to count the 7s

3.10 Let’s play a game, how many 7s can you count?

3.11 Let’s play a game, how many 7s can you count?

3.12 Let’s play a game, how many 7s can you count?

3.13 Visual perceptions

Our perception of data on a typical printed page is associated with several visual variables.
We call them aesthetics

3.14 Data visualisations

The use of data visualisations can improve:

Ease of understanding
Engagement and attention
Efficient communication
Enhanced memory retention
Cross-Disciplinary Understanding
Presentation flexibility

3.15 Data visualisations in Sport

Visualisations can be used to enhance:

Performance Analysis
In-Game Insights
Player Tracking and Biometrics
Fan Engagement
Scouting and Recruitment
Tactical Analysis
Predictive Analytics

3.16 From data to visualisation

3.17 Aesthetics

Aesthetics are quantifiable features within a graphic.
Aesthetics can take different forms.
The type of data we are working with will determine which aesthetic (or combination of aesthetics) we can best use.

3.18 Aesthetics

3.19 Task

Map your variable onto a relevant aesthetic

3.20 Aesthetics

Not all aesthetics can represent continues data
- E.g. shape and line type cannot be used for continues data
Discrete data such as categorical data (ordered or unordered), text, or quantitative discrete variables (e.g. scale 1-5) can be represented by most aesthetics.

3.21 Using aesthetics

Data should be mapped onto aesthetics
Creating a scale
When creating a scale each unique value needs to have a unique aesthetic value
- Reason why shape and line type cannot be used on continues data
Often visualisations use three scales, however it is possible to have more than 3 scales in one visualisation

3.22 Task

Design an imaginary sport visualisation with 3 different scales. Can you make one with 4 different scales?

3.23 Coordinate systems and axes

Visualising data requires position scales
Most commonly used system is Cartesian coordinate system
- x and y coordinates
Each grid spacing on the x- and y-axis refers to a step in the variable unit (e.g. 1 league point).
X- and y-axis with different units don’t require same spacing.
- Stretch along y-axis to emphasis on y-axis change
- Stretch along x-axis to emphasis on x-axis change
If x and y use same units, spacing should be equal to not distort your message.

3.24 Coordinate systems and axes

Non-linear axis are not uncommon.
- Log-transformed – often used when variables have a very different magnitude.
- Square root – less often used but may be useful when your data contains 0’s.
When using log-transformation ensure you are clear when plotting the data.
Another commonly used coordinate system in sport data analytics is the polar system.
- X-axis is circular

3.25 Using colour

Distinguish
- i.e. make difference between groups/ categorical data clear

3.26 Using colour

Represent
- i.e. use colour to show the value of continues variables

3.27 Using colour

Highlight
- i.e. focus on one specific group or element within your data

3.28 Types of visualisations

Lots of different types of visualisations available.
Which one to use depends on the data you are displaying
- Amounts
- Distribution
- Proportions
- Relationships
- Geospatial data
- Uncertainty

3.29 Visualising amounts

Visualising amounts refers to visualising a value for a set of categories (e.g. Olympic medals per country)
Bar plots most commonly used
- “Normal”, stacked, grouped
- Pay attention to labelling
- Ordering
Dot plots and heatmaps are alternative options

3.30 Examples of amounts

3.31 Visualising distributions

Visualising distributions refers to visualising the relative proportions of different variables.
Histogram and density plots most common
- You need to set bins (histogram) or bandwidth (density) – arbitrary.
- Note density plots can display data that does not exist, be aware of this
Alternatives empirical cumulative distribution function (ecdf) and quantile-quantile plots (q-q plots)

3.32 Empirical cumulative distribution function

ECDF ranks all data points based on value from small to large (or vice versa).
To increase readability and information the y-axis is often normalized to the maximum rank so the maximum y-value equals 1.

3.33 Empirical cumulative distribution function

3.34 Quantile-quantile plots

q-q plots are a useful when we want to determine to what extent the observed data points follow a given distribution.
q-q plots use ranks to predict where a given data point should fall if the data were distributed according to a specified reference distribution (often a normal distribution).

3.35 Visualising proportions

3.36 Visualising proportions

3.37 Visualising proportions

3.38 Visualising proportions

3.39 Visualising associations

Scatter plots
Correlation diagrams
Slope graphs