At the end of 2011, the Institute of Advanced Architecture of Catalonia organized an international competition entitled ‘ City sense: shaping with real-time data ‘. Architecture, which has long been outside the debate on this emerging issue, is also coming to be achieved by the ‘ Data ‘ virus. Data-Atia ecosystem is old like the world but, when called “data”, it is that it lends itself to different numerical experiments. Among them, data mining is based on the observation that a growing and already gigantic mass of data of all types is available. It reminds us that the issue now lies less in the collection of these data than in their treatment and in their use to make them meaningful. In the case of predictive data analysis-one of the modes of data mining-the idea is that the meaning to be searched is no longer in each data as such but in the results of massive correlations within a data ecosystem. Wikipedia defines the predictive analysis of data as encompassing “a variety of techniques derived from statistics, data extraction and game theory that analyze facts present and past to make predictive assumptions about Future events “. Each has experienced the predictive analysis of data, observing, after entering keywords in a search engine, the appearance of proposals for complementary keywords and associated responses. When Google, to name but one, claims to specify your search in your place, it is the result of a predictive analysis of your data, articulating the ones you are providing and those you have produced in your previous Visits. Predictive data analysis also makes it more complex to predict epidemics from cross-medical site consultations with geographic data, to assess the risk associated with an insurance contract from data Describing the context of the operation and the previous behaviour of the insured, classifying financial behaviour in a normal or suspicious category, among others *. Closer to our concerns of architects and urban planners, conclusive results are beginning to emerge in the field of predictive modeling of land use in intensive agricultural context on the basis of winter cover of soils * *, Prediction of urban floods and real-time management of floods, anticipation of the attendance of cultural establishments-by crossing the modeling of the flows of foreigners arriving in France with the profile of tourists, which determines their system of choice of places of visit-. A paradigm of anticipation the predictive data analysis postulates, therefore, that an adequate numerical treatment allows to reveal a field of probable more or less precise and therefore to anticipate. The key concept of predictive data analysis is indeed that of anticipation, as opposed to deduction. Anticipation is in a probabilistic logic, while the deduction is anchored in a deterministic logic. Anticipation-etymologically “ante capex” i.e. taking the lead-admits and assumes a margin of error. It has a dose of hypotheses and arbitrary choices, even fiction (anticipatory novel is not a novel of deduction). This is basically a tool that almost superimposes the past states, the present state and sketches of future states. In other words, it provides an instantaneous synthesis between a diagnosis and a project form. and urban planning? The predictive analysis of data is a question: 1/massive collection of datas, 2/algorithmic processing and probabilistic of these. On the first point, the city is undoubtedly a gigantic source of data: more or less fine demographic statistics, economic, social statistics, cadastral data, building permit data and work declarations, followed by Waste collection and sorting, monitoring of energy consumption and production, consumption and treatment of water… This information is updated in more or less long space-time, varying from several years to the week or even to the day or, for some, in real time. In the field of production of the city, they are treated by different bodies in studies with a more or less prospective vocation: Local planning plan, study of movements, local plan of the habitat, territorial coherence schemes, urban projects Various. Other data are captured and processed in real time, such as bus movements and their speed of travel, allowing them to predict their time limits for switching to different stops. This approach is practiced rather by those who act a priori only marginally on the structure of the city, considering it as a medium. At the same time, everyone produces almost continuous, consciously or not, voluntarily or not, an unimaginable mass of geolocalized data: via its Internet searches, its phone or its smartphone with GPS, via the use of applications Various like Coyote ® or Tripadvisor, via Twitter or Facebook, using public transport, tolls, via surveillance cameras… These data are produced and collected in real time or in a slight delay. Sometimes scientists take it but they are so far exploited mainly by private companies, for commercial purposes. It is the phenomenon of ‘ crowdsourcing ‘ or ‘ supplying information to crowds ‘, which is constantly developing and finding ever more diverse applications. In summary, the city secretes data collected at long intervals and used to act on the urban structure in depth, data collected in real time and used for the different uses of the city, without changing the “matter” and a mass of Data collected in real time but almost unused. So the city of data exists. Remains to be defined-to imagine? -the “City of Predictive data analysis“.

Claire Bacon

* EPF-4th year-IPA-data Mining-4 Courses: Modeling-page 6/16-Bertrand Liaudet * * PhD thesis in Geography by Samuel Corgne, Costel UMR CNRS 6554 LETG in cotrusteeship with ENST-Bretagne, University of Rennes 2-Upper Brittany references: 1. We-are-data.com: Works of the students of the ESAG directed by Bernard kissed and Pierre-Yvon Carnoy 2. ‘ Place de la Canvas ‘, Xavier de la Porte, issue of September 4, 2011