Existing Problem and Our Approach Abstract ...

Which Individuals make the best Team? Designing an Analytics Platform for Professional Sports Teams #sports #soccer #wearable_sensors #team_analysis #event_stream_processing #big_data_innovation #icis_2015

This research study has been published and presented at the International Conference on Information Systems 2015 in Fort Worth, TX, USA. The full article can be downloaded under: http://aisel.aisnet.org/icis2015/proceedings/ISdesign/1/

Overall Research Project

Design Principles

Joint Big Data Innovation Lab Step 1

Step 2

Step 3

Interviews were recorded, transcribed, and stepwise coded. In particular, codes were developed to identify abstract design principles for a professional sports teams platform. Based on these design principles, we developed and iteratively refined a prototype.

Step 4

SAP, University of Mannheim, and Karlsruhe Institute of Technology

Design Principles

Fawad A. Shah, Martin Kretzer, Alexander Mädche

Existing Problem and Our Approach

Explore potentials of wearable sensors for professional team sports

Propose design principles to address the identified potentials and build a prototype

• Problem: Team sports statistics are usually computed by aggregating individual team member statistics. However, oftentimes not the best individual team members make the best team. • Our goal: Design a Team Sports Analytics Platform for measuring performance of groups of players • Technical challenge: Very detailed data is required to compute good team KPIs. Thus, players are equipped with wearable sensors. These need to be analyzed using event stream computing.

Abstract Professional sports is an area with huge economical and societal impact. However, so far it has received rather little attention by information systems research. Therefore, we design an Analytics Platform for Professional Sports Teams. We introduce Sports Analytics as an interdisciplinary research field and establish a design science research project. As part of this project, we first explore three professional sports teams and theorize meta-requirements for our platform. Second, we propose design principles that indicate how the proposed metarequirements may be addressed. Third, we develop a prototypical web application that indicates how the proposed design principles may be instantiated. Finally, we evaluate the application with two professional sports teams and a global software vendor. As part of this evaluation, we focus on relevance and feasibility of the proposed Analytics Platform for Professional Sports Teams.

Install sensors at training fields and collect data

1. In order to support analysis of team performance, a Team Sports Analytics Platform should provide team-dependent KPIs, i.e., playerinterplay KPIs, in addition to individual player KPIs. 2. In order to support tactical decisions, a Team Sports Analytics Platform should allow definition of specific tactics and prioritization of KPIs according to those tactics. 3. In order to support different tactics with different combinations of players, a Team Sports Analytics Platform should support definition of groups of players. The number of players per group should be flexible. 4. In order to support team lineup decisions, a Team Sports Analytics Platform should allow comparison of alternative groups of players. 5. In order to link video material with granular computations, a Team Sports Analytics Platform should link data from a media server (for video data) and a four-tier architecture consisting of data collection, data preprocessing, data storage and processing, and data presentation (for lowlevel positional data from sensors attached to players’ wear and the ball.

Evaluate whether the identified potentials could be realized

Interviews We conducted 15 interviews with participants from three different, professional (men’s) soccer teams. Participants included coaches, players, specific athletic coaches and doctors. We conducted semistructured in-depth interviews to gain detailed, rich, and real-life data.

Prototype Team

Team Type

Interviews

Deutscher Fussball Bund (German National Team)

National

2

TSG 1899 Hoffenheim (highest German league)

Club

11

FC Astoria Walldorf (4th highest German league)

Club

2

DP2 DP1

DP4

DP3