Sensing Vehicle
 
An average car today is equipped with over 4,000 sensors that collect internal and external data that make the car run. Now, for the first time, researchers at MIT, in collaboration with Volkswagen Group of America Electronic Research Lab, have used this data for another purpose: to analyze driver behavior and the urban environment. The results of their research are published in the Proceedings of the IEEE, and could help reduce car accidents, alleviate driver stress, and provide us with a better sense of how drivers interact with their environment. Over the past few years, cars have been transformed from the kinds of mechanical systems Henry Ford might have imagined into veritable computers on wheels, filled with thousands of sensors. We asked the question: what could we extract from this wealth of information? Could we use it to better understand how drivers make decisions, and to improve overall safety on the roads and in our cities?





Urban Lens
http://senseable.mit.edu/urban-lens/
Urban Lens explores millions of anonymized financial transactions in Spain based on data provided by BBVA, which holds a ubiquitous banking infrastructure in the country. The data provides an opportunity to uncover macro trends derived from a fine-grained scale of individual economic behavior. In light of the failure of past decades to produce models that effectively predict and explain the macroeconomic trends, we noticed that a gap exists between models of micro behaviors and macro phenomena. This project performs a comparative analysis of city microeconomics, aiming to elucidate how bigger economic patterns could be understood utilizing data of individual economic transactions. We built a novel multi-scale predictive model of Spanish regions, quantifying the distinctive signature of each region based on their spending behavior by identifying indicators regarding the amount of spending, type of spending, type of individual, and individual mobility. The model was validated at the provincial scale using official performance statistics, and proved a strong correlation between individual spending behavior and official socioeconomic indices. Finally, a scale-free clustering was developed to enable a consistent aggregation of regions in different spatial dimensions.