Working with data in elite sport

class: bottom, center, title-slide

# Working with data in elite sport
### Dr Jacquie Tran   |   <span class="citation">@jacquietran</span> |   29 Nov 2019

---

class: right, middle
background-image: url(https://raw.githubusercontent.com/jacquietran/2019_stats_teachers_day/master/images/uofabg.png)
background-size: cover

## Today's session

Intro to performance sport and sport science

Explore how data is used in performance sport

Work together on a sports data problem

---
class: inverse, center, middle

# A bit about me...

---
background-image: url(https://raw.githubusercontent.com/jacquietran/2019_may_rladies_akl/master/images/aus_map_base.jpg)
background-size: cover

.footnote[
Image credit: [**University of Melbourne**](https://biomedicalsciences.unimelb.edu.au/departments/pharmacology/engage/avru/discover/snakes/common-brown-snake)
]

---
class: center, top
background-image: url(https://raw.githubusercontent.com/jacquietran/2019_may_rladies_akl/master/images/eastern_brown_snake_aus.jpg)
background-size: cover

.footnote[
Image credit: [**University of Melbourne**](https://biomedicalsciences.unimelb.edu.au/departments/pharmacology/engage/avru/discover/snakes/common-brown-snake)
]

---
background-image: url(https://raw.githubusercontent.com/jacquietran/2019_may_rladies_akl/master/images/flinders_st.jpg)
background-size: cover

.footnote[
Image credit: [**Flickr**](https://www.flickr.com/photos/neelelora/6987389739/)
]

---
class: inverse, center, middle

# "Applied sport science"

# 🤔

---
background-image: url(https://raw.githubusercontent.com/jacquietran/2019_may_rladies_akl/master/images/deakin_sprint_start.jpg)
background-size: cover

.footnote[
Image credit: **Deakin University**
]

---
background-image: url(https://raw.githubusercontent.com/jacquietran/2019_may_rladies_akl/master/images/deakin_gait_lab.jpg)
background-size: contain

.footnote[
Image credit: **Deakin University**
]

---
background-image: url(https://raw.githubusercontent.com/jacquietran/2019_may_rladies_akl/master/images/deakin_vo2.jpg)
background-size: cover

.footnote[
Image credit: **Deakin University**
]

---
background-image: url(https://raw.githubusercontent.com/jacquietran/2019_may_rladies_akl/master/images/deakin_gym.jpg)
background-size: contain

.footnote[
Image credit: **Deakin University**
]

---
background-image: url(https://raw.githubusercontent.com/jacquietran/2019_may_rladies_akl/master/images/Match_Analysis_Portable_K2_Panoramic_Video_Camera_System.jpg)
background-size: contain

.footnote[
Image credit: [**Wikimedia**](https://en.wikipedia.org/wiki/File:Match_Analysis_Portable_K2_Panoramic_Video_Camera_System.jpg)
]

---
background-image: url(https://raw.githubusercontent.com/jacquietran/2019_may_rladies_akl/master/images/UniNutrition-043-1.jpg)
background-size: contain

.footnote[
Image credit: [**University of Bath**](https://www.teambath.com/physio-sport-science/sports-nutrition/)
]

---
background-image: url(https://raw.githubusercontent.com/jacquietran/2019_may_rladies_akl/master/images/Sports-Psychology1.jpg)
background-size: 85% 85%

.footnote[
Image credit: [**Boxing News**](http://www.boxingnewsonline.net/how-to-use-sports-psychology/)
]

---
class: inverse, center, middle

# A potted history of data in sport

---
background-image: url(https://raw.githubusercontent.com/jacquietran/2019_may_rladies_akl/master/images/question-mark.jpg)
background-size: cover

## For many years...

Sports performance data has been **challenging** to collect.

<br />
<br />
<br />
Imagine that you want to measure how fast an athlete sprints over a short distance.

**How would you do this?**

---
class: center
background-image: url(https://raw.githubusercontent.com/jacquietran/2019_may_rladies_akl/master/images/av_hill_speed_testing.png)
background-size: 65% 65%

## A.V. Hill in 1927

.footnote[
Image credit: [**Bassett, 2002, J Appl Physiol**](https://www.semanticscholar.org/paper/Scientific-contributions-of-A.-V.-Hill%3A-exercise-Bassett/fce9096c04e4425f30ba6ebe78a026c6b3be2ea6)
]

---
class: center
background-image: url(https://raw.githubusercontent.com/jacquietran/2019_may_rladies_akl/master/images/deakin_light_gates.jpg)
background-size: 60% 65%

## A contemporary solution, with **LASERS**

.footnote[
Image credit: **Deakin University**
]

---
class: center

## Sports analytics today

We have more data than we know what to do with!

---
class: center

## Sports analytics today

We need to (learn to) work accurately and efficiently with high-resolution data.

---

## A general workflow for sports analytics

**Determine the need** through collaboration

Articulate the need as a **question**

Scope out the **"minimum viable product"**

Allow time for **peer review**

**Communicate insights** in appropriate ways

---
class: inverse, center, middle

# Sports analytics use cases

## Example 1: Mining text data

---
class: center

## Knowledge Edge for Tokyo

Cross-sport, cross-time evidence

Surveys and interviews

Repeated data collection

---
class: inverse, center, middle

# 🤐

---
class: inverse, center, bottom

## The Sportswoman's Library, Vol. II (1898)

---

.pull-left[
![](https://raw.githubusercontent.com/jacquietran/2019_may_rladies_akl/master/images/illus-310.jpg)
]

.pull-right[
*"For any form of outdoor exercise, the two chief requisites of costume are warmth and lightness. A thin flannel shirt is more useful than anything, worn with a short light skirt."*
]

.footnote[
Image credit: [Project Gutenberg](https://www.gutenberg.org/files/47243/47243-h/47243-h.htm#LAWN-TENNIS)
]

---
background-image: url(https://raw.githubusercontent.com/jacquietran/2019_may_rladies_akl/master/images/sportswomans_lib_frequency.png)
background-size: 85% 80%

---
background-image: url(https://raw.githubusercontent.com/jacquietran/2019_may_rladies_akl/master/images/sportswomans_lib_bigrams.png)
background-size: 85% 80%

---
class: inverse, center, middle

# Sports analytics use cases

## Example 2: Team scoring dynamics

---
class: center, middle

![](https://raw.githubusercontent.com/jacquietran/2019_may_rladies_akl/master/images/merritt_clauset.PNG)

[**Merritt & Clauset, 2014**](https://link.springer.com/article/10.1140/epjds29), *EPJ Data Science*

---
background-image: url(https://raw.githubusercontent.com/jacquietran/2019_may_rladies_akl/master/images/merritt_clauset_fig3.PNG)
background-size: contain

---
background-image: url(https://raw.githubusercontent.com/jacquietran/2019_may_rladies_akl/master/images/afl_tables_home.PNG)
background-size: cover

---
background-image: url(https://raw.githubusercontent.com/jacquietran/2019_may_rladies_akl/master/images/afl_tables_score_progression.PNG)
background-size: contain

---
class: center, middle

![](https://raw.githubusercontent.com/jacquietran/2019_may_rladies_akl/master/images/tran_letter_header.PNG)

<a href="https://www.jsams.org/article/S1440-2440(17)31300-2/abstract" target="_blank">**Tran & Letter, 2017**</a>, *J Sci Med Sport*

---
class: center, middle
background-image: url(https://raw.githubusercontent.com/jacquietran/2019_may_rladies_akl/master/images/tran_letter_1.png)
background-size: contain

---
class: center, middle
background-image: url(https://raw.githubusercontent.com/jacquietran/2019_may_rladies_akl/master/images/tran_letter_2.png)
background-size: contain

---
class: center, middle
background-image: url(https://raw.githubusercontent.com/jacquietran/2019_may_rladies_akl/master/images/tran_letter_3.png)
background-size: contain

---
class: inverse, center, middle

# Sports analytics use cases

## Example 3: Possession chains

---
class: center, middle

![](https://raw.githubusercontent.com/jacquietran/2019_may_rladies_akl/master/images/di_domenico_soccer.PNG)

---
class: center, middle

---
background-image: url(https://raw.githubusercontent.com/jacquietran/2019_may_rladies_akl/master/images/di_domenico_rationale.png)
background-size: contain

---
background-image: url(https://raw.githubusercontent.com/jacquietran/2019_may_rladies_akl/master/images/di_domenico_header.png)
background-size: contain

---
background-image: url(https://raw.githubusercontent.com/jacquietran/2019_may_rladies_akl/master/images/di_domenico_results_turnovers.png)
background-size: 70% 90%

---
class: inverse, center, middle

# Let's work with some sports data!

---

<br />
This exercise uses a publicly available data set that includes **all podium results from the Winter Olympic Games from 1924 to 2014, inclusive**.

The `winter` data set is downloadable from this link:  
**[https://www.kaggle.com/the-guardian/olympic-games/data](https://www.kaggle.com/the-guardian/olympic-games/data)**

---
class: inverse, center, middle

# "Start with the end in mind."

---

## Considering different perspectives

What makes the cut as a **"motivating question"**?

It depends on **perspective**, **priorities**, and **roles**.

---
class: center

## The athlete

---
class: center

## The coach

---
class: center

## The performance analyst

---
class: center

## The performance director

---

## Activity

Imagine you are in one of these roles: athlete, coach, performance analyst, or performance director. Think: **What matters to me in this role?**

---

## Activity

Think about and / or note down **your initial questions** about the data set we're working with, which contains all podium results from the Winter Olympic Games from 1924 to 2014, inclusive.

---

## Activity

**Discuss** in small groups, then we'll share with the wider group.

- I am an / a [ athlete | coach | analyst | performance director ]
- What matters to me is ...
- The kinds of questions I am interested in are ...

---

## A worked example

I am a **performance director** for a Winter Olympics sport in NZ.

What matters to me is **keeping an eye on competitor countries** to benchmark ourselves and learn from what our opposition does well.

The kinds of question I am interested in are **about how other countries have performed** at the Winter Olympics, especially those that have **favourable environmental conditions** for training and competing.

---
class: center, middle

## **How many gold medals** were won by
## **Canada, Norway, and Sweden** at the
## **last five Winter Olympic Games**, up to 2014?

---

## Subset the data

To answer this question, we create a data subset that focuses only on:

- Gold medal results

- Athletes from Canada (CAN), Norway (NOR), or Sweden (SWE), and

- Results from the five Winter Olympics between 1998 to 2014, inclusive.

---
class: middle

<table class="table" style="margin-left: auto; margin-right: auto;">
 <thead>
  <tr>
   <th style="text-align:right;"> Year </th>
   <th style="text-align:left;"> City </th>
   <th style="text-align:left;"> Sport </th>
   <th style="text-align:left;"> Discipline </th>
   <th style="text-align:left;"> Athlete </th>
   <th style="text-align:left;"> Country </th>
   <th style="text-align:left;"> Gender </th>
   <th style="text-align:left;"> Event </th>
   <th style="text-align:left;"> Medal </th>
  </tr>
 </thead>
<tbody>
  <tr>
   <td style="text-align:right;"> 1998 </td>
   <td style="text-align:left;"> Nagano </td>
   <td style="text-align:left;"> Biathlon </td>
   <td style="text-align:left;"> Biathlon </td>
   <td style="text-align:left;"> BJOERNDALEN, Ole Einar </td>
   <td style="text-align:left;"> NOR </td>
   <td style="text-align:left;"> Men </td>
   <td style="text-align:left;"> 10KM </td>
   <td style="text-align:left;"> Gold </td>
  </tr>
  <tr>
   <td style="text-align:right;"> 1998 </td>
   <td style="text-align:left;"> Nagano </td>
   <td style="text-align:left;"> Biathlon </td>
   <td style="text-align:left;"> Biathlon </td>
   <td style="text-align:left;"> HANEVOLD, Halvard </td>
   <td style="text-align:left;"> NOR </td>
   <td style="text-align:left;"> Men </td>
   <td style="text-align:left;"> 20KM </td>
   <td style="text-align:left;"> Gold </td>
  </tr>
  <tr>
   <td style="text-align:right;"> 1998 </td>
   <td style="text-align:left;"> Nagano </td>
   <td style="text-align:left;"> Bobsleigh </td>
   <td style="text-align:left;"> Bobsleigh </td>
   <td style="text-align:left;"> LUEDERS, Pierre </td>
   <td style="text-align:left;"> CAN </td>
   <td style="text-align:left;"> Men </td>
   <td style="text-align:left;"> Two-Man </td>
   <td style="text-align:left;"> Gold </td>
  </tr>
  <tr>
   <td style="text-align:right;"> 1998 </td>
   <td style="text-align:left;"> Nagano </td>
   <td style="text-align:left;"> Bobsleigh </td>
   <td style="text-align:left;"> Bobsleigh </td>
   <td style="text-align:left;"> MACEACHERN, David </td>
   <td style="text-align:left;"> CAN </td>
   <td style="text-align:left;"> Men </td>
   <td style="text-align:left;"> Two-Man </td>
   <td style="text-align:left;"> Gold </td>
  </tr>
  <tr>
   <td style="text-align:right;"> 1998 </td>
   <td style="text-align:left;"> Nagano </td>
   <td style="text-align:left;"> Curling </td>
   <td style="text-align:left;"> Curling </td>
   <td style="text-align:left;"> BETKER, Jan </td>
   <td style="text-align:left;"> CAN </td>
   <td style="text-align:left;"> Women </td>
   <td style="text-align:left;"> Curling </td>
   <td style="text-align:left;"> Gold </td>
  </tr>
  <tr>
   <td style="text-align:right;"> 1998 </td>
   <td style="text-align:left;"> Nagano </td>
   <td style="text-align:left;"> Curling </td>
   <td style="text-align:left;"> Curling </td>
   <td style="text-align:left;"> FORD, Atina </td>
   <td style="text-align:left;"> CAN </td>
   <td style="text-align:left;"> Women </td>
   <td style="text-align:left;"> Curling </td>
   <td style="text-align:left;"> Gold </td>
  </tr>
  <tr>
   <td style="text-align:right;"> 1998 </td>
   <td style="text-align:left;"> Nagano </td>
   <td style="text-align:left;"> Curling </td>
   <td style="text-align:left;"> Curling </td>
   <td style="text-align:left;"> GUDEREIT, Marcia </td>
   <td style="text-align:left;"> CAN </td>
   <td style="text-align:left;"> Women </td>
   <td style="text-align:left;"> Curling </td>
   <td style="text-align:left;"> Gold </td>
  </tr>
  <tr>
   <td style="text-align:right;"> 1998 </td>
   <td style="text-align:left;"> Nagano </td>
   <td style="text-align:left;"> Curling </td>
   <td style="text-align:left;"> Curling </td>
   <td style="text-align:left;"> MCCUSKER, Joan </td>
   <td style="text-align:left;"> CAN </td>
   <td style="text-align:left;"> Women </td>
   <td style="text-align:left;"> Curling </td>
   <td style="text-align:left;"> Gold </td>
  </tr>
</tbody>
</table>

---

## Wrangle the data

In the `winter` data set, the data is structured such that **one row is one medal-winning athlete, per country, per year**.

However...

- Team events (e.g., bobsleigh, curling) comprise multiple athletes, and

- Within teams that achieve a podium finish, each athlete is awarded a medal.

For this analysis, we need to wrangle the data to get it into a format where one row represents **one gold medal per event, per country, per year**.

---

## Create an 'identifier' variable

We can use these existing variables to create a unique ID for each gold medal won per country per event:

- Year
- Sport
- Discipline
- Country
- Gender
- Event

---

## Check the new variable

<table class="table" style="margin-left: auto; margin-right: auto;">
 <thead>
  <tr>
   <th style="text-align:left;"> unique_event_ID </th>
  </tr>
 </thead>
<tbody>
  <tr>
   <td style="text-align:left;"> 1998_Biathlon_Biathlon_NOR_Men_10KM </td>
  </tr>
  <tr>
   <td style="text-align:left;"> 1998_Biathlon_Biathlon_NOR_Men_20KM </td>
  </tr>
  <tr>
   <td style="text-align:left;"> 1998_Bobsleigh_Bobsleigh_CAN_Men_Two-Man </td>
  </tr>
  <tr>
   <td style="text-align:left;"> 1998_Bobsleigh_Bobsleigh_CAN_Men_Two-Man </td>
  </tr>
  <tr>
   <td style="text-align:left;"> 1998_Curling_Curling_CAN_Women_Curling </td>
  </tr>
  <tr>
   <td style="text-align:left;"> 1998_Curling_Curling_CAN_Women_Curling </td>
  </tr>
  <tr>
   <td style="text-align:left;"> 1998_Curling_Curling_CAN_Women_Curling </td>
  </tr>
  <tr>
   <td style="text-align:left;"> 1998_Curling_Curling_CAN_Women_Curling </td>
  </tr>
  <tr>
   <td style="text-align:left;"> 1998_Curling_Curling_CAN_Women_Curling </td>
  </tr>
</tbody>
</table>

---

## Identify and omit rows with duplicate IDs

...so that one row = one gold medal per event, per country, per year.

---
class: center, middle

Now we have tidy data in the format we need for analysis!

---

## Calculate gold medal totals

We use the wrangled data to calculate the **total number of gold medals** won by **Canada, Norway, and Sweden** at each of the **Winter Games between 1998 and 2014**.

---

## Calculate gold medal totals

From this...

---

## Calculate gold medal totals

...to this:

---

Go beyond tables and make visual interpretation easier!

![](https://raw.githubusercontent.com/jacquietran/2019_essa_forum/master/images/demo1_data_plot.png)

---
class: inverse, center, middle

# Sports and recreation data sets

---

## Tennis

.pull-left[
<img src="https://raw.githubusercontent.com/jacquietran/2019_stats_teachers_day/master/images/ausopen.jpg" width="550px" />
]

.pull-right[
Open Era (1968 to now) Grand Slam finals results

[**(Women's Singles link)**](https://data.world/fsd01/tennis-grand-slam-championships-champion-vs-runner-up-women)

[**(Men's Singles link)**](https://data.world/fsd01/tennis-grand-slam-championships-champion-vs-runner-up-men)
]
---

## Football (Soccer)

.pull-left[
<img src="https://raw.githubusercontent.com/jacquietran/2019_stats_teachers_day/master/images/epl.JPG" width="550px" />
]

.pull-right[
English Premier League game statistics over 10 seasons from 2009/10 to now

[**(link)**](https://datahub.io/sports-data/english-premier-league)
]
---

## Cricket

.pull-left[
<img src="https://raw.githubusercontent.com/jacquietran/2019_stats_teachers_day/master/images/whiteferns.JPG" width="550px" />
]

.pull-right[
Ball-by-ball information for men's and women's cricket across all formats

[**(link)**](https://cricsheet.org/)
]

---

## Bike-sharing

.pull-left[
<img src="https://raw.githubusercontent.com/jacquietran/2019_stats_teachers_day/master/images/bikeshare.JPG" width="550px" />
]

.pull-right[
New York City Citi Bike trip data

[**(link)**](https://www.citibikenyc.com/system-data)
]

---
class: inverse, center, middle

## @jacquietran