Social Data Analysis&physical Activity

Social data analysis & physical activity Olivier Liechti

Abstract

Institute for Information and

In this paper, we describe the uBike on-line service, primarily designed for amateur cyclists. We explain how the collective analysis of data is used to promote physical activity. The data fed into the system consists of GPS tracks captured during training sessions. Data analysis provides the basis for a range of features, including personal progress reports, group rankings and match-making of potential training partners.

Communication Technologies (IICT) University of Applied Sciences of Western Switzerland (HEIG-VD) Av. des Sports 20 1401 Yverdon-les-Bains [email protected] Christophe Burnet Institute for Information and

Keywords

Communication Technologies (IICT)

Sport, ubiquitous computing, GPS tracking, cycling, Web2.0, physical activity, wellness, health, augmented communities, collective data analysis

University of Applied Sciences of Western Switzerland (HEIG-VD) Av. des Sports 20 1401 Yverdon-les-Bains

Introduction

[email protected]

The lack of physical exercise is an increasing concern in industrialized countries and there is a clear need to raise public awareness. To tackle this problem, some approaches are merely informational (e.g. communication campaigns from government agencies). Other approaches are economical (e.g. incentives given by health insurance companies). Yet other approaches seek to make physical activity more enjoyable. The reasoning is that people may become more active if they do not perceive physical activity as a burden or if they get some psychological reward out of it. We strongly believe in this idea. One goal of the uBike project is to create a testbed for research in this area.

Copyright is held by the author/owner(s). CHI 2008, April 5 – April 10, 2008, Florence, Italy ACM 1-xxxxxxxxxxxxxxxxxx.

Of course, technology is not required to make physical activity enjoyable. For instance, playful activities can be designed for children without the need for cutting-edge equipment. Nevertheless, we argue that it is worth exploring the relationship between ubiquitous computing technologies, social interaction and sport [1, 2, 3, 4, 5]. Video games that make physical movements an integral part of the gaming experience [6] have already shown some success. They are an example of an entertaining technology that has positive side effects on wellness. The uBike service was designed in this context. It is best described as a "Web2.0" service for the community of amateur cyclists. Its concept is very simple and can be summarized in three points. Firstly, data is recorded during training sessions with a GPS tracker. Secondly, information is extracted from the raw data. Finally, a range of features that exploit the information are provided to users. What makes the service interesting, is that the data is not only analyzed on an individual basis, but collectively. GPS data is used to foster social interactions between users, both on-line and in the real world. This is one of our strategies to make exercise more appealing. In the remaining paragraphs of this paper, we first look at some of the barriers to physical activity. We then give an overview of the uBike service and describe some of its features. We briefly describe the user interface and provide some snapshots. Finally, we describe the current status of the project and outline future work.

Overcoming barriers to physical activity Several studies [7, 8] have looked at the reasons cited by people to explain their lack of physical activity. From the following list of ten reasons, we have selected five: 

Do not have enough time to exercise



Find it inconvenient to exercise



Lack self-motivation



Do not find exercise enjoyable



Find exercise boring



Lack confidence in their ability to be physically active (low self-efficacy)



Fear being injured or have been injured recently



Lack self-management skills, such as the ability to set personal goals, monitor progress, or reward progress toward such goals



Lack encouragement, support, or companionship from family and friends, and



Do not have parks, sidewalks, bicycle trails, or safe and pleasant walking paths convenient to their homes or offices.

These arguments have raised several interesting questions during the design of the uBike service. Firstly, can we make exercise more enjoyable and less boring? It is true that "suffering" alone on a bicycle does not seem particularly appealing. Unlike soccer or tennis, cycling is not a game – it is pure effort. But we can think of at least two ideas. The first idea is to find a way, so that people do not have to exercise alone. A mechanism that allows potential training partners to identify each other is a step in this direction. The second idea is to to add a competitive dimension to amateur cycling. Maintaining rankings of "fastest

racers", "most courageous cyclists" or "most regular trainers" allows a user to compare himself with others. The progression in these rankings certainly provides a motivation to train on a regular basis. Secondly, can we help people monitor their performances and their progress towards personal goals? This is an obvious feature of the system: the data provided by a user is processed and personal statistics are computed over time. Yet, interesting questions can be explored here. How should the information be presented to users? Showing numbers and graphs is certainly useful, but one could also think

Figure 1. Screenshots of the uBike service

of more compelling and engaging interaction techniques. For instance, we plan to integrate dynamically generated narratives to draw the attention of users (e.g. "Congratulations, you have established a new record", "Keep going, you are about to reach the 2'000 km milestone", etc.). The last question, is whether we can we address the "lack of companionship" cited by many people? Designing uBike as a community-oriented service, with a range of collaborative features is an attempt to do it. We make the distinction between group-oriented features and community-oriented features [9]. Users

explicitly join groups (typically because they already know some other users), whereas they are implicitly part of the larger community. Within a group competition is a strong motivation factor. Within the community, real-life encounters are a motivation factor.

Overview of the uBike service The uBike service is a web application that has a lot in common with other popular "Web 2.0" services. Users can log into the system, upload data and access various features. Some of these features are "social" in the sense that they foster interactions between users. Compared to other services, the uBike does not only seek to foster on-line interactions. One goal of the system is to also foster interactions in the real world, in the form of joint training sessions. We use the term "augmented community" to stress that we aim to combine on-line and real-world interactions, which both have their benefits and drawbacks [9].

out that storage capacity is actually a more serious problem. Many devices have a limited capacity, which means that recorded data has to be transferred after every single trip. For this reason, we have decided to use trackers with external memory cards for the upcoming pilot study. This might seem like a detail, but end-to-end usability is important for the acceptance and success of the service. Phase 2: data analysis Once GPS files have been transferred to the uBike service, they are analyzed by the processing engine. We have designed this sub-system in way that allows the creation of "processing chains" by assembling a series of plug-ins. This makes the system extensible and will allow us to easily add new features, in an iterative fashion. Currently, we have developed plug-ins for the following tasks:

Usage of the service can be summarized with three phases: 1) data collection, 2) data analysis and 3) data exploration and exploitation.



Trip segmentation and merging. One file uploaded to the server may contain several trips, and one trip may be split in several files. Furthermore, files may arrive out-of-order on the server.

Phase 1: data collection The data used by the service is collected during training sessions. Cyclists carry a basic GPS device that records track points. When they get back home, the GPS data is transferred to a PC and then uploaded to the uBike web site. In other words, data is processed off-line and does not have to be transmitted in realtime.



Feature extraction. Once a trip has been extracted from a set of files, features such as the distance, the time, the elevation, the maximum speed, the average speed and the bounding box can be computed and recorded.



Route identification. In our terminology, a trip is an instance of a particular route. If a user is cycling 5 times from Town A to Town B, we are dealing with 5 trips and 1 route. It is useful to link trips and routes, because it supports progress monitoring (e.g. the user has established a new record for a particular route).

During the first project phase, we have evaluated a number of off-the-shelf GPS trackers. We expected that battery life was going to be a concern, since we did not want to bother users and hinder acceptance. It turned





Computation of statistics. Features extracted from the GPS tracks are used to compute and maintain various statistics. Different time scales are considered, because it is interesting to monitor progress in the short term (how am I doing this week?), in the medium term (what is my progress since the beginning of the season?) and in the long term (how am I doing compared to previous years?). Graph generation. Visualization of collected data is important, because we want the service to be informative but also entertaining and engaging. In the processing chain, static graphs can be generated (birds-eye view of a trip, profile views of a trip, statistical bar charts, etc.). In the future, we also plan to add interactive visualization features (e.g. virtual races).

Phase 3: data exploration & exploitation Once GPS tracks have been processed, information is available and can be explored by means of various features: progress reports, rankings, graphs, etc. As shown in Figure 1, we have segmented uBike features into three categories: i) individual features (available on the "My Data" tab), ii) group features (available on the "My Buddies" tab) and iii) community features (available on the "My Community" tab). An integration with Google Maps has also been realized. Users within a given region are displayed on the map. Information about their training habits and fitness level can be gathered.

Current status & future work One goal of the uBike project was to design, build and deploy a system in order to make experiments with a significant user base. We are progressing towards this goal. In Spring 2007, the concept was formalized and a mock-up was designed. It was used to gather feedback from prospective users. The implementation phase was then started. At the time of writing, the first release is almost complete. Our goal is to have a functional system in production for the next cycling season. We will focus on individual features first, until a critical mass has been reached. We plan to gradually add more features and to asses their effectiveness.

References

[1] Monitoring, measuring and motivating exercise: ubiquitous computing to support fitness, Ubicomp 2005 workshop, 11-14 Sept. 2005, Tokyo. [2] E. H. Chi, G. Borriello, G. Hunt, and N. Davies. Pervasive computing in sports technologies. IEEE Pervasive Computing, 4(3):22-25, 2005. [3] J. F. McCarthy and T. D. Anagnost.MusicFX: An Arbiter of Group Preferences for Computer Supported Collaborative Workouts, CSCW '98, 14-18 Nov. 1998, Seattle. [4] MotionBased, http://www.motionbased.com [5] http://www.apple.com/ipod/nike/ [6] Nitendo Wii, http://wii.nintendo.com/ [7] JF Sallis, MF Hovell. Determinants of exercise behavior. Exercise and Sport Science Reviews 1990;18:307-330. [8] http://www.cdc.gov/nccdphp/dnpa/physical/life/over come.htm [9] O. Liechti. Fun on the wheels: the uBike social application. Ubiwell workshop, Ubicomp 2007, 16 September 2007, Insbruck, Austria.