Quantified Self (Predictive Analytics)

 

The quantified-self movement has existed for many years as a collaboration of people collecting continual data on their everyday activities in order to make better choices about their health and behaviour. But, with today’s Internet of Things, the movement has begun to come into its own and have a wider impact.
Smartphones contain a rich record of people’s activities, including who they know (contact lists, social networking apps), who they talk to (call logs, text logs, e-mails), where they go (GPS, Wi-Fi, and geotagged photos) and what they do (apps we use, accelerometer data). Using this data, and specialized machine-learning algorithms, detailed and predictive models about people and their behaviours can be built to help with urban planning, personalized medicine, sustainability and medical diagnosis.
For example, a team at Carnegie Mellon University has been looking at how to use smartphone data to predict the onset of depression by modelling changes in sleep behaviours and social relationships over time. In another example, the Livehoods project, large quantities of geotagged data created by people’s smartphones (using software such as Instagram and Foursquare) and crawled from the Web have allowed researchers to understand the patterns of movement through urban spaces.
In recent years, sensors have become cheap and increasingly ubiquitous as more manufacturers include them in their products to understand consumer behaviour and avoid the need for expensive market research. For example, cars can record every aspect of a person’s driving habits, and this information can be shown in smartphone apps or used as big data in urban planning or traffic management. As the trend continues towards extensive data gathering to track every aspect of people’s lives, the challenge becomes how to use this information optimally, and how to reconcile it with privacy and other social concerns.