Linked articles

Open data: why and how to standardize datasets?

Since October 7, 2016, open data has become the rule for all actors invested with a public service mission with more than 50 agents and for territories with more than 3,500 inhabitants. "While the law imposes a principle of generalized openness, each territory publishes data according to its competencies, its data assets and its practices," Data Publica observes in a very comprehensive file devoted to the standardization of open data.This booklet presents a comprehensive overview of the standardization of open data: it focuses on the challenges related to the design and reuse of open data in local governments.After recalling the growing importance of standardization in information technology, and highlighting the considerable standardization work that underlies the production of statistics, the authors of the booklet retrace the numerous initiatives in favor of open data standardization. They look in detail at cases of sectoral standards (mobility data, in particular), point out the challenges of designing standards (issues of consultation, the difficulty of choosing the "right" format), and then those of producing standardized data (standardizing data that has already been published, supporting the production of new data), before concluding with a discussion of alternatives to standardization.Open data differ from one territory to another"From one producer to another, the files do not necessarily contain the same fields or give the same level of detail," the authors of this booklet observe. "The values in the fields themselves are not standardized.Data are not named in the same way in different territories. " In addition to differences in terminology between communities, there is a more general mismatch between producers, who publish documents with their own vocabulary, and users who formulate their needs with another."In concrete terms, these issues of discoverability and data standardization limit the impact of open data. Without harmonization of practices, it is very complicated to build services or uses that go beyond a single territory. The authors of the report take as an example the Handimap application, which proposes accessible routes for people with reduced mobility by taking into account low sidewalks, and which has been hampered in its development by the lack of standardization of data on road accessibility. Each new local instance of the application required significant development to adapt to the local data.Standards allow different digital tools to communicate to build a coherent whole in the service of one or more objectives. Data standardization could thus reduce friction " by facilitating the discovery of similar open data in different territories and by making it possible to consolidate locally produced data into a national database that can be easily exploited. Standardize data with schemas"Open data standardization is built around schemas. These are machine-readable standards, conventions that describe the fields and values allowed in a dataset that conforms to its recommendations. It is therefore by conforming to them that we produce standardized data sets. Machine-readable, the schemas are re-used in forms and interfaces for human use.Following the example of the world of statistics, where the comparability of indicators is based on their standardization, the world of geographic and environmental information has been engaged in advanced work on standardization since 2007. The INSPIRE directive proposed provisions for the interoperability of geographic and environmental data through the standardization of metadata but also of the data itself. In France, the Conseil National de l'Information Géolocalisée (CNIG) is the main developer of geographic data standards (generally regulatory standards). In 2022, the former "data" commission was renamed "standards" commission, attesting to the important place held by the design of geostandards in the activities of the CNIG.Beyond geographic data, France has recently seen a resurgence of interest and a proliferation of initiatives on the issue of standardization.The number of data schemas governing the production and reuse of data, particularly by local authorities, is increasing.Since 2018, the OpenDataFrance association, which federates local authorities committed to an open data approach, has been developing the Socle Commun des Données Locales (SCDL) to homogenize the open data publication of essential data produced by territorial actors, to help producers improve the quality of the data they publish, and to facilitate the exploitation of published data by reusers.Eight datasets previously selected as priorities have been standardized and the base is being extended.The SCDL has also spurred momentum in the state administration with the June 2019 launch of schema.data.gouv.fr, which references French standards that have been adopted by regulation or designed by the community of data producers and reusers. The site also references schemas that are under investigation and construction. Schema.data.gouv.fr: produce schemas collaboratively to homogenize dataIn the summer of 2022, nearly fifty data schemas are referenced, including a dozen under investigation. The documentation proposed on the site allows data producers to appropriate these schemas and thus to produce data sets that respect them.Following the Prime Minister's circular of April 27, 2021, on the public policy of data, algorithms and source codes, 15 ministerial roadmaps were published on September 27, 2021.The Ministry of Territorial Cohesion's roadmap sets the objective of "encouraging the opening of data according to shared reference systems is a guarantee of quality that, in the long run, will facilitate interoperability and even the emergence of open solutions. In collaboration with associations of local authorities, pioneering territories at different scales, as well as publishers of digital solutions equipping local authorities, it is a matter of converging and promoting the best practices of standardization" (Action15).

• Data

• Commons

Published on 02/02/23



Ten years of open public data: open data begins a new phase

At the very beginning of the 2010s, a few public actors voluntarily engaged in the opening of public data, mainly metropolises.On February 21, 2011, the decree creating a "Mission Etalab" was published. Its role: to create the single interministerial portal for public data: data.gouv.fr10 years later, the data.gouv.fr platform has given rise to numerous sectoral variations) as well as API.gouv.fr and data sharing.The recognition, in 2016, by the Law for a Digital Republic of the principle of open public data as a matter of principle marked an important step. From now on, open data is no longer just a choice of a few, it has become an obligation for all actors invested with a public service mission.Open data has played a central role in crisis management to inform citizens and steer public policies. Public interest in open data and its uses has probably never been so strong.Companies are getting involved in the subject, as users of the data of course, but also sometimes by opening their own data.Following the report by Deputy Eric Bothorel on public policy on data, algorithms and source code, the Prime Minister reminded the ministries and regional prefects on April 21, 2021, that data policy is "a strategic priority for the State in its relations with all its partners, particularly local authorities and private actors. He invites the various ministries to appoint a ministerial data administrator by May 15.New challenges, new ambitions: the opening of data is entering a new phase. Covid and public data: a wealth of dashboards and visualization tools to track the pandemicAt the beginning of March 2020, the only available data on the epidemic were scattered in the press releases of the ARS and Santé Publique France. These data were fragmentary and did not allow to draw curves and to follow the evolution of the epidemic.It is a group of citizens who will take care of it. In a few days, a collective (the "OpenCOVID19" initiative) and then a community of 200 data scientists organized themselves to manually extract data from 20 ARS and 100 prefectures and integrate them into a spreadsheet. A very heavy load: everything has to be redone every day[1]. The veille-coronavirus.fr dashboard ( takes shape and offers a first consolidated vision of the official data. The implementation of this contributory tool quickly receives the support of Etalab developers, then the support of the Ministry of Health. On March 28, Édouard Philippe and Olivier Véran, in a press conference, used the infographics of veille-coronavirus.fr.As of April 2020, 177 Covid-19-related datasets were online at data.gouv.fr.Since March 18, Santé Publique France publishes daily updated datasets on the public data platform data.gouv.fr. and on Geodes, the mapping observatory of Santé Publique France.This opening of data has fed the work of researchers and opened the way to a myriad of visualizations in the press. The datavisualisations of Gouvernement.fr and dashboards such as Covidtracker, Coronaboard.fr, Covinfo.fr or Vaccinator.fr rely on Geodes data.At the beginning of March 2021, SPF proposed 123 indicators on its Geodes website, at all scales (national, regional, departmental, territorial). Launch of the 10th spring of data.gouv.frOn the occasion of its 10 years of existence, Etalab is organizing a whole series of events to take stock of its actions, its successes and failures ... and to imagine the next 10 years. It is a dialogue that we want to launch with you."We are well aware that the public digital landscape has changed significantly over the past decade. New questions are emerging, which also call for new answers: how can we collectively use data to serve the general interest? How can we take advantage of artificial intelligence without sacrificing ethics and public and individual freedoms? What are the profiles that the public service must acquire, today and tomorrow, to pursue its digital transformation? What can we learn from Etalab's counterparts in Europe and around the world?" The month of April was dedicated to data quality. May is the month of data reuse. The month of June will be dedicated to data discoverability. "The data policy, a strategic priority of the State in its relations with all its partnersIn a circular dated April 27, 2021, addressed to ministers and regional prefects, the Prime Minister reminds us that administrations "must constantly seek the best circulation of data, algorithms and codes, in open formats that can be used by third parties. This renewed ambition implies, in addition, a strengthening of the opening of source codes and public algorithms, as well as the use of free and open software, and the extinction, by 2023, of fees charged for the reuse of data. It invites ministers to designate a ministerial data administrator by May 15 and regional prefects to designate a "data, algorithms and source codes" referent.Finally, with regard to relations with private actors holding so-called "general interest" data with high added value for the public authorities, he announced the forthcoming creation of a general interest data mediator.An interdepartmental strategy for the digital sector including skills related to data and algorithms will be proposed by the Minister of Public Transformation and Civil Service by May 15, 2021.The objectives of opening up the data and codes of the ministries must be translated into a roadmap finalized by July 15 and published by September 15. This roadmap will have to display the priorities of the ministries in terms of opening up data and codes and include a section on training agents and managers in data issues.Administrations are also encouraged to systematically reference their data on data.gouv.fr or api.gouv.fr, as Dinum is responsible for improving the quality of government portals. The site code.gouv.fr, listing software and algorithms opened by administrations, will soon complete the system. Open Data Impact: nine challenges for the next decade of open public dataThe Open Data Impact program, led by the Fing, has undertaken a critical assessment of these two years of open public data. "What is open data today? What have been the impacts of the actions carried out over the last eight years? What has been successful with open data? Have the promises been fulfilled?"In a first step, the authors of this report focused on "factualizing the diagnosis": "open data, "how many divisions? To date, France has more than 40,000 published data sets (more than 35,000 listed on data.gouv.fr). In number, more than three quarters come from local authorities, which essentially publish small and medium-sized data sets. The State publishes comparatively fewer games, in number, but, by nature, large quantities in volume, especially through large reference games with national coverage.  A thematic view of the data shows a very varied set. Some areas are particularly well covered: geography, economy, transportation, etc. There are few blind spots or data-poor areas: social issues, justice and, to a lesser extent, education or culture. "Some portals," note the authors of this report, "judiciously propose to users, for each dataset, to publish the results of their uses. For example, such and such a dataset has allowed the creation of such and such an application. This practice is clearly of interest to enhance the value of uses and to promote more and more future uses. The number of reuses of each set gives an indicator, admittedly partial, but revealing of the demand for certain sets. The data.gouv.fr platform provides a list of the games that have been used the most. Among the datasets that have been reused, a large third are long-standing and traditionally used government data, such as INSEE data, the postal code database, or certain data specific to the work and operations of major ministries: Interior (election results, accidentology), Health (Finess, doctor demographics, medicines), Economy and Finance (Fantoir, Land Values)... "Open data has further democratized their use and the published reuses show new uses of these data. There are also "many data previously limited to certain audiences, which Open Data has brought out of the shadows. The fuel price database is an emblematic example: previously sold for several tens of thousands of euros a year, its release in Open Data has developed its uses. Finally, for a quarter, this list highlights the rapid success of datasets directly born of the Open Data dynamic, such as the National Address Base (BAN), a project driven by Etalab, involving the IGN, La Poste and OpenStreetMap. What are the impacts? In a second step, the authors of the report review five major families of impacts of open public data: "economic impacts: has open data created a market for the reuse of data? can this market be evaluated quantitatively? which private economic actors have taken up the subject? democratic impacts: has open data strengthened the transparency of public life? is it a tool for monitoring the actors of public life (elected officials, agents of administrations, etc.) and/or a tool for empowering citizens? the impacts on organizations that have opened data or use them, and their relationships with third parties: has open data enabled the establishment of relationships with new ecosystems? How does open data change existing relationships, for example with public service providers? the impact on jobs, functions and working methods: which jobs are mainly impacted by the opening of data? Are we seeing the emergence of new functions, or even new jobs? Has open data changed working methods within organizations? Finally, the impacts on the diffusion of a data culture within organizations: is open data the first brick of data literacy? Has open data paved the way for topics like the smart city?" Nine challenges In conclusion, the authors of the report identify nine challenges for the next decade of open public data, classified into three main themes: to make people see / want to use open data, to develop an offer that is truly adapted to the demand and to reexamine the boundaries of open data.Making people see, making them want to take advantage of open dataChallenge1: finding relevant metrics. How can we "measure" what the opening of public data produces? The two types of metrics used today (the number of reuses and the measurement of the audience of open data portals) are insufficient to capture everything that open data produces. It is time to imagine and test other metrics.Challenge 2: Open data, the pillar of intelligent cities and territories, should finally be explained and experienced by the greatest number of people. How can we explain the uses? How can we facilitate them, make them even more accessible? Many actors are trying, partially, in a scattered manner. Portals are gradually integrating more services. This challenge proposes to make it easier to see, to make people want to see and to facilitate the transition to action. Developing a data offer that is truly adapted to demandChallenge3: make data truly reusable. Is it necessary to repeat that preparation represents 80% of the time spent using open data? And that a considerable number of datasets are hardly reusable. Do we need to repeat that without standardization, without quality of service, a regular professional use of open data is not possible? Ease of use of data, one of the greatest challenges of open data today?Challenge 4: for a demand-driven open data! The energy devoted to opening a maximum of data has probably reached its limits. Just look at the many sets that have never been used. Can we imagine, on the contrary, an open data driven by demand? by what we already know to be successful? How can we identify and listen to this demand? How can we respond to it under the right conditions?Challenge 5: How can we encourage producers to make a lasting commitment, other than by constraint? One possible pitfall of introducing the principle of open data into the law would be to lose the meaning of this obligation for public actors. What is at stake here is the long-term incentive for producers to adopt an open approach, because it seems to us that this is one of the keys to the subject for the next decade. Re-users expect more than just compliance with legal obligations, they also expect a real commitment (to improve the data, to ensure the sustainability of the offer, to make an effort to improve quality, etc.).Re-examine the boundaries of open dataChallenge6: Tracking use without hindering it? Making use conditional on identification is one of the initial yellow lines of open data. But is it possible to track usage more precisely and systematically without hindering it? Should we follow the users ... or the uses?Challenge 7: From data to information. The notion of "raw data" was imposed very early on as the primary condition for "open data". However, this notion is fragile and we have known for a long time that data has necessarily undergone transformations before being published. Going further, we wonder about the possibility of publishing information, easier to understand and interpret, in addition to the so-called raw data. We can probably go even further by publishing targeted information, refined, adapted to practical and immediate uses, etc. But this does not call into question the value of raw data.Challenge 8: Open data without platforms or applications. Can we be satisfied with open data portals? Is open data always intended to produce an application? We believe that other forms of open data can emerge, outside of traditional platforms and uses.Challenge 9: Itchy data: for an open data that is not harmless! The pioneers of open data were intuitively cautious, and courageous publications and "itchy" uses are rather rare. How can we encourage them? Which data are problematic? Gender equality, the fight against discrimination, the fight against climate change: how can data usefully contribute to the major debates in society? European Union merges portals to promote public data sharingTo date, the European Union has two Open Data portals. The European Data Portal collects the metadata of public data available on the Open Data portals of the EU countries. The EU Open Data Portal makes available the datasets of the EU institutions and bodies.In the spring of 2021, Europe will merge these two currently separate gateways into a single Open Data site. This new portal will give access to more than one million datasets, from 36 countries, 6 institutions and 79 European bodies and agencies. IGN public consultation on the co-production and opening of geo-commons"In France as well as abroad, more and more public authorities are seizing the dynamics of the "commons" as a third dimension to the public and the private. Initiatives mobilizing communities on the ground outside of any structure, including associations, are multiplying. At the same time, geolocalized data have taken over our daily lives and have given rise to multiple services and applications. Thanks to smartphones and GPS, geographic information has become widely available.This context offers a more open and collaborative playground for the institute: the geo-commons.After the opening and free access to most of the IGN's data since January 1, 2021, the institute wishes to be more broadly involved in this dynamic of building commonalities and to build, with its ecosystem, its new strategy around the concept of geo-commons: a set of geographic information databases (production) and digital tools (dissemination) that are accessible to the greatest number of people.The online public consultation " IGN and the Commons " invites professionals, users of geographic information, local authorities and partner institutions to participate in the collective impetus to build the geo-commons

• Data

Published on 12/05/21



Opening up content and data in cultural institutions: where do we stand?

88% of cultural institutions in France (libraries, archives and museums) are digitizing all or part of the works and document collections they hold and 76% are putting them online, according to a survey conducted by Wikimedia France and the Phare agency among 145 libraries, archives and museums.In this sense, the policy of opening up data driven by the Ministry of Culture has borne fruit: a policy defined in 2013policy, extended and formalized in 2021 in its "Roadmap - Cultural Data and Content"..However, the Wikimedia France survey draws attention to the fact that only one third (31%) of institutions adopt "open licenses".Wikimedia France is thus reopening the debate on the restrictions on the reuse of digital cultural content.At the heart of this debate is the notion ofOpen Content, which " characterizes specific and distinct approaches to Open Data or Open Access based on two objectives:"To facilitate the opening, putting online and making available of cultural content on the one hand ;Facilitate the optimal reuse of these cultural contents based on a number of parameters (choice of license, size and quality of images, etc.)".For the promoters of Open Content, this implies not only putting cultural and heritage content online and making it available, but also making it possible to reuse it, including commercially, by adopting open licenses such as CC-BY or Etalab.9 out of 10 cultural institutions digitize contentAmong the digital practices examined in this survey, the digitization of cultural content is the most mentioned.In total, nearly 9 out of 10 cultural institutions (88%) say they have already digitized content from their collections. Of these, 72% say that this is a regular activity and 16% that it is more occasional."If cultural institutions, in general, are very far from having digitized all of their collections and works (...) there is now a profusion of collections digitized by cultural institutions, and therefore a potential reservoir of works accessible and reusable by the public. This potential, very often, is unevenly developed because cultural institutions present a great disparity of means and human resources".The study highlights an increasing relationship between the level of digitization adoption and the size of the institution: "the larger the institution, the more frequent digitization practices are observed".While 55% of institutions with 2 to 5 employees report regularly digitizing their content, this ratio rises to 80% for institutions with 15 to 50 employees and 100% for institutions with more than 50 and 99 employees.66% of cultural institutions use social mediaCultural institutions have been relying on social networks (Twitter, Facebook, Instagram) for a few years now to promote content and reach new audiences.66% of institutions report using it regularly. 19% only occasionally. Only 15% of respondents do not report using these tools or did not answer the question.This practice is more frequent in museums (80%) and libraries (71%) than in archives (48%).56% say they frequently post images from their collections onlineAmong the practices of putting images or heritage content online (also called deposits), the survey distinguishes between putting textual data (records) about the collections online and putting images from the collections online.35% of the institutions declare that they regularly post data on their collections (23% sometimes and 38% never) and 56% post images from their collections.Among the institutions that put images of their collections online, 61% do so on dedicated portals, 15% on external sites.These practices of uploading images to a dedicated portal or to an external portal, far from being exclusive, are complementary. "They reflect the attention of a significant number of cultural institutions in France to make it possible to consult the images in their collections through a variety of channels.Conditions for reuse of images from the collections are still heterogeneous"The use of licenses to define the conditions for making images available online by cultural institutions is frequent, but not systematic.While 41% of the institutions declare to put images of their collections online using licenses, only 26% of the respondents declare to use free licenses (such as CC-BY or Etalab);The choice of a particular license varies according to the type of cultural institution: free licenses (such as CC-BY) are mostly adopted by libraries (18%) and museums (12%), less frequently by archives (4%);"The proportion of institutions involved in Open Content approaches, defined by the use of licenses (such as CC-BY or Etalab open licenses) allowing free and open reuse - for commercial and non-commercial purposes - remains relatively low," conclude the authors of the study. Lower in museums than in libraries. And lower than the practices observed in our European neighbors. Opening up content in cultural cultural institutions: a global trendSince the beginning of the 2010s, museums, libraries and archives have been deploying practices of opening and sharing their content without specific restrictions for users.Three institutions in particular are international references in this field and still inspire many institutions in these approaches today:The Rijksmuseum in Amsterdam, as of 2012, has given free access and unrestricted permission for high definition images of these collections;The Getty Museum, in 2013 made reproductions of works in its collections available online allowing for commercial and non-commercial use;The Metropolitan Museum of New York made a portion of its collections available online for non-commercial use in 2014 and then in 2017 decided to make images of public domain works more widely available for any unrestricted reuse.While in 2016, only 50 institutions were engaged in these approaches, that number jumps to 324 in 2018."For many of these cultural and heritage institutions, an Open Content approach is an opportunity to broaden the dissemination of their content to a much wider, transnational if not globalized audience. However, practices can vary greatly from one institution to another (restriction of commercial uses or not, etc.). Grenoble, Paris, Rennes, Toulouse, Marseille... Announcements of the opening of cultural data in France have been multiplying over the past three years. Since 2020, Paris Musées has offered Open Content (free and unrestricted access) to more than 150,000 digital reproductions of works from the collections of the City of Paris museums in high definition. In 2020, the City of Marseille has adopted the principle of opening up more than 200,000 heritage resources (writings, objects and works) already digitized by the city.

• Data

Published on 18/11/22



Potential impacts of open data reuse: use cases to discover

"What problems are we addressing? What are the proposed solutions? How is open data being used? What are the impacts?"  The Etalab Mission, which runs the data.gouv.fr platform, is keen to publicize the diversity of uses and potential impacts of open data. selection of examples The purpose of this project is to present a series of examples of uses that highlight the diversity of subjects covered, the possible impacts and the people who use open data."Whether it is visualizations, applications, press articles, research papers, etc., the reuse of data can promote the transparency of the State, the transformation of public action, the information of citizens or the creation of new services. The call of the emergency services : Locate precisely the interventions Our Green Cities : Knowing the tree heritage of the municipalities Weak Signals Identifying companies in difficulty Property Value Data (PVD) Knowing the sale price of real estate Ekylibre : Simply manage a farm DiagOriente Offer an orientation course ERRIAL : Carry out the risk assessment of a property BeeGIS Estimating the floral resources around the hives

• Data

• Innovation

Published on 12/09/22



Local government acts: online publication becomes the norm from July

Provided for by an ordinance and a decree published last October, the overhaul of the procedures concerning the publicity and entry into force of local acts will essentially apply as of next month.Localtis, the news feed of the Banque des Territoires, describes the implementation of these new obligations as a "small revolution" for local authorities. A "small revolution". for local authorities. List of the deliberations of the commune From July 1, 2022, the acts of municipalities with more than 3,500 inhabitants, but also of EPCIs, departments and regions, will no longer have to be published in paper format but in electronic form, on the website of the local authority or institution.The minutes of the meetings of municipal councils and of the deliberative bodies of inter-municipalities will be abolished as of July 1. Municipalities and inter-municipalities will have to draw up a list of the deliberations of their deliberative body, which will have to be posted at the headquarters of the community or grouping, and published on its website - when it exists - within one week.The communes and their groupings will always be required to draw up minutes of the meetings of local deliberative assemblies. The document will be published in electronic form when the commune or grouping has a website. Dematerialized publication One of the major novelties brought about by the reform is precisely related to dematerialization: local authorities will no longer be obliged to post or publish their acts in paper format. As of July 1, regulatory acts and decisions of the kind - that is, acts that are neither regulatory nor individual - must be published in electronic format.In other words, electronic publication will constitute "the common law formality of publication".However, the dematerialization will not be imposed on municipalities of less than 3,500 inhabitants. Six-month deadline for urban planning documents As of January 1, 2023, "all local authorities and their groupings, without exception", will have to publish their planning documents on the national planning portal.In order to help the communities concerned to prepare for this small revolution, the collectivites-locales.gouv.fr website publishes a dedicated page, a dozen or so sheets.

• Innovation

Published on 15/07/22



Données et quartiers : comment mettre la science des données au service des quartiers prioritaires de la ville ?

Pendant 3 ans, l’association RésO Villes a entrepris de jeter une passerelles « entre deux univers qui ne se connaissaient pas : celui de la politique de la ville et celui de la data ».Avec le soutien de l’Agence Nationale de la Cohésion des Territoires (ANCT) et en s’appuyant sur les experts de CIVITEO, le programme « Data & Quartiers » a expérimenté de nouveaux outils et de nouveaux usages de la donnée que l’on retrouve plus fréquemment déployés dans les quartiers centraux au titre des projets de « ville intelligente ».Les objectifs principaux du programme étaient les suivants :▪ Utiliser des outils de « datascience » au bénéfice des quartiers prioritaires ;▪ Améliorer la connaissance et l’observation des quartiers grâce à la donnée ;▪ Comprendre et expliquer ces nouveaux outils, ces nouvelles données, ces nouveaux usages ;▪ Documenter la méthodologie pour favoriser la duplication des expérimentations.Les projets menés sur les thématiques de l’emploi, de la santé et des mobilités, avec le soutien de partenaires privés et publics, ont permis de consolider une méthodologie et d’acquérir certaines convictions quant au rôle de la donnée dans la politique de la ville. Les objectifs initiaux du programme « Data & Quartiers »Utiliser des outils de la science des données dans les quartiersEn 2018, les Quartiers Prioritaires ne font pas partie des terrains d’expérimentations de ces nouveaux usages. RésO Villes a fait le choix d’engager une réflexion sur l’utilité des outils de la science des données pour résoudre des problématiques spécifiques des quartiers. L’enjeu central du programme va être de mobiliser des données disponibles à l’échelle des villes, quelle qu’en soit la source, pour proposer de nouvelles méthodes d’observation des quartiers. Les outils utilisés dans des quartiers « intelligents » des centres villes arrivent dans les quartiers périphériques.Améliorer la connaissance et l’observation des quartiers grâce à la donnéeIl existe des indicateurs de suivi des quartiers, notamment auprès de l’Insee et de l’Observatoire national de la politique de la ville. Ces indicateurs permettent de suivre des statistiques utiles : démographie, emploi, pauvreté… Mais cette approche ne permet pas de connaître et de comprendre la vie quotidienne des habitant.e.s. Et pourtant des données existent, elles sont même de plus en plus nombreuses.Documenter la méthodologie pour favoriser la duplication des expérimentationsEn 2018, les projets recourant à la science des données ne sont pas très fréquents chez les acteurs publics, de surcroît à l’échelle de deux régions et dans une démarche partenariale associant des acteurs privés. Pour ces raisons, les partenaires de RésO Villes ont très vite insisté sur la nécessité de documenter les processus. Leur engagement dans le programme visait autant à comprendre (et à apprendre) la méthode qu’à accéder à des résultats inédits.Tous les acteurs du programme ont été sensibilisés à cette démarche, y compris les prestataires réalisant des travaux de science des données. Il leur a constamment été demandé de documenter non seulement les méthodes et les outils retenus, mais aussi ceux qui ont été testés et laissés de côté.Expliquer ces nouveaux outils, ces nouvelles données, ces nouveaux usages, ces nouveaux enjeux« Data & Quartiers » avait un objectif plus large l’acculturation et la sensibilisation des différents acteurs du secteur public en général, et de la politique de la ville en particulier, sur les sujets de la donnée. À ce titre, diverses actions ont été menées auprès de publics variés ayant des niveaux de maturité numérique très disparates. Entre 2019 et 2022, de nombreux événements ont été organisés : conférences (à l’attention des conseillers citoyens de Bretagne et des Pays de la Loire, des responsables politique de la ville et/ou numériques des collectivités partenaires), conférences ouvertes à des acteurs territoriaux de la France entière organisées par l’Agence Nationale de la Cohésion des Territoires et le Centre National de la Fonction Publique Territoriale, des retours d’expérience auprès de différents centres de ressources politiques de la ville (Occitanie, PACA, Centre Val-de-Loire...) La Méthodologie « Data & Quartiers »Dès le départ, le programme « Data & Quartiers » a retenu un protocole en cinq étapes.Étape n°1 : sensibiliserL’étape de sensibilisation proposée à tous les acteurs du programme avait deux objectifs :un objectif général d’acculturation : la donnée est aujourd’hui au cœur de tous nos actes personnels, en tant que consommateur.rice.s, usager.e.s mais aussi citoyen.ne.s ;un objectif ciblé : définir ensemble les conditions du partage et de la mise à disposition de données en travaillant en commun les finalités des expérimentations à venir.La sensibilisation dissipe les craintes éventuelles mais permet aussi de fixer les limites. L’acculturation peut se faire de différentes façons : une formation « théorique » présentant les enjeux, des ateliers de mise en situation des participants ou des séquences en ligne de type webinaire.Étape n°2 : cadrer les besoins et définir les « cas d’usage » de la donnéePour chaque sujet, quel que soit le thème :Partir d’un constat ou d’une réalité. Débuter d’une photographie de la réalité éprouvée dans les quartiers prioritaires. Elle est fondée sur la connaissance du terrain des experts de la politique de la ville. À ce stade, il n’est pas encore question de donnée…Formuler une hypothèse. Une hypothèse est formulée pour améliorer la situation décrite préalablement. On commence à aborder la donnée comme un levier de résolution des difficultés rencontrées. On formule par exemple l’hypothèse que des données existent pour améliorer la connaissance des habitudes des habitant.e.s des quartiers (des données pour appréhender les difficultés d’accès à l’emploi, pour comprendre la prévalence de problématiques de santé, pour connaître les habitudes d’alimentation, pour connaître les pratiques culturelles…).Poser une problématique. La problématique prend la forme d’une question précise. C’est l’objet des travaux. Concise et directe, elle décrit les ambitions fixées par le groupe de travail. Par exemple : les habitant.e.s des quartiers sont-ils plus exposé.e.s que d’autres au Covid-19 ?Définir des objectifs. Les objectifs constituent la feuille de route de chaque groupe de travail : identifier des données, créer des conditions pour y accéder (s’il ne s’agit pas « d’open data »), les collecter, les croiser, produire de la visualisation des données…Identifier les données utiles: Chacun des projets du programme s’est appuyé sur une ou plusieurs séances d’exploration de données, que l’on appelle aussi le « datamining ». Cet exercice est bien connu des expert.e.s de la donnée, mais il est peu utilisé pour les politiques publiques locales, et encore moins à l’échelle des quartiers.Étape n°3 : valider le cas d’usage / la preuve de conceptAccéder aux données. Au fil des mois, l’équipe de « Data & Quartiers » a développé une certaine expertise pour organiser l’accès aux données. Il y a d’un côté les données librement accessibles, et notamment les données en « open data » mises à disposition par l’Insee ou publiées par les communes. Mais il y a aussi d’autres données publiques non accessibles. Il s’agit de données produites par des acteurs publics, mais non rendues publiques. Il y a aussi des données privées. Elles sont produites par des entreprises qui acceptent de les mettre à disposition de RésO Villes au nom de l’intérêt général.Préparer et explorer les données. Une fois les données collectées, il convient de s’assurer qu’elles sont exploitables au regard des objectifs. Les experts de la donnée prennent alors la main et vont explorer les différents fichiers et les bases de données mises à disposition. Ils vont notamment se pencher sur leur qualité : le format et le type de données, leur granularité (c’est-à-dire leur précision et leur fréquence de mise à jour), leur distribution et leur complétude (« a-t-on bien toutes les données utiles ? »), leur qualité (données aberrantes, incohérences…), etc. Cette phase peut donner lieu à des allers-retours entre les « fournisseurs » et les « explorateurs » de données afin de les enrichir et de les compléter.Modéliser: En fonction des données disponibles et des objectifs à atteindre, les expert.e.s de la donnée vont appliquer une ou plusieurs techniques de modélisation. Il peut s’agir de modèles statistiques classiques ou d’algorithmes et de techniques plus avancés, y compris ayant recours à l’intelligence artificielle. Les expert.e.s de la donnée vont ainsi tester les jeux de données et voir si les informations que l’on espère en tirer vont en découler.Évaluer le modèle. Les expert.e.s de la donnée appliquent leurs modèles sur les jeux de données de tests afin d’en comparer les performances. Les résultats obtenus sont ensuite présentés aux « experts métiers » afin d’en valider la pertinence. Parfois, il faut corriger les choses, par exemple en collectant des données différentes. Ce n’est qu’une fois la démarche validée, que l’on se lance dans un traitement à grande échelle des données. Appliquée à « Data & Quartiers », cette méthode a par exemple conduit à travailler sur 5 ou 6 quartiers sur des données de santé, avant d’étendre le travail aux 78 QPV de Bretagne et des Pays de la Loire.Étape n°4 : prototyper la solutionUne fois vérifiée la pertinence des données et démontrée l’utilité de la modélisation (avec ou sans intelligence artificielle), il faut réfléchir à la manière de rendre les données « actionnables », c’est-à-dire trouver une forme de traitement des données qui va permettre aux acteurs de s’en servir de façon concrète. De nombreuses questions se posent : faut-il un outil que chacun peut manipuler à distance ? Faut-il une représentation statique (un tableau, une carte…) ? Comment représenter des données complexes à l’échelle de chaque quartier ? A l’échelle des villes ou des régions ? Les données doivent-elles être figées dans le temps ou faut-il prévoir de les actualiser ? Avec quel degré d’automatisation ?Étape n°5 : mettre en productionLe programme « Data & Quartiers » dans sa version 2019-2022 était un programme expérimental. Il s’est arrêté à l’élaboration des prototypes. Mais l’objectif logique d’une démarche de science des données est de fournir ensuite des outils qui seront intégrés dans les processus de travail au quotidien. Trois projets mis en œuvre dans le cadre de « Data & Quartiers »Données et Emploi dans les quartiers : une étude sur les distances domicile-travailLe groupe de travail Emploi s’est fixé comme objectif d’étudier la réalité de la distance entre les métiers recherchés par les habitant.e.s des quartiers « politique de la ville » et les emplois à pourvoir qui y correspondent. Ce travail est réalisé par les équipes de science des données du groupe Randstad. Les données utiles à la résolution de cette problématique ont été transmises par Pôle emploi. Il s’agit des données des métiers les plus recherchés par les habitant.e.s des quartiers, d’après les déclarations effectuées lors des premiers entretiens avec les agents de Pôle emploi. Bien sûr, ces données sont anonymes. Elles ont ensuite été croisées avec les données des offres d’emploi à pourvoir, fournies par Randstad. Les cartographies produites permettent de visualiser la distance entre les emplois les plus recherchés par les habitants des quartiers et la localisation des offres d’emploi.Données et Santé dans les quartiers : une étude sur la fragilité des habitants face au Covid-19Différents travaux semblaient montrer que toutes les populations ne sont pas égales face au Covid-19, que certaines présentent des risques plus importants que d’autres d’être contaminées ou de développer des formes graves, et puisque des facteurs sanitaires ou sociaux étaient annoncés comme pouvant être à l’origine de cette inégalité, RésO Villes a souhaité objectiver la réalité des risques en Bretagne et Pays de la Loire.Le premier travail a été de sélectionner des indicateurs explicatifs d’une certaine fragilité face au virus : des indicateurs sanitaires et des indicateurs sociaux. Les indicateurs sociaux renvoient aux conditions de vie ou sociales des habitant.e.s des quartiers et permettent de cibler géographiquement ces habitant.e.s. L’avantage de ces données c’est qu’elles sont facilement accessibles, du moins à l’échelle des communes ou des IRIS. Des données existent sur le taux de personnes bénéficiaires de la CMU-C (devenue C2S), ou encore sur le taux de population vivant en logement HLM, ces deux éléments combinés révélant un niveau de ressources et un mode d’habitat Les données sanitaires été transmises par l’Observatoire régional de la santé de Bretagne.L’équipe en charge des travaux de géomatique a livré dans un premier temps des cartographies statiques sur un nombre restreint de territoires. Puis un outil plus sophistiqué a été développé.Les résultats des travaux valident à l’échelle locale les études nationales et internationales les ayant inspirés : toutes les populations ne sont pas égales face au Covid-19 et certaines présentent des risques plus importants d’être contaminés et/ou de développer des formes graves.Pour la Bretagne, les résultats sont éloquents : les quartiers prioritaires de la ville (QPV) sont systématiquement situés au cœur des zones dans lesquelles le risque de fragilité face au Covid-19 est le plus fort (même si toutes les zones fragiles ne sont pas des QPV).Données et Mobilités dans les quartiers : une étude sur les modalités de déplacement Domicile-EmploiLa distance domicile-travail opérée par les habitant.e.s des QPV ne peut représenter à elle seule, un critère déterminant dans la recherche d’un emploi. Identifier des distances entre domicile et emploi est une chose. Savoir comment s’y rendre en est une autre. Une convention de mise à disposition de données a d’abord été signée entre RésO Villes et l’Université de Rennes 2 pour permettre la réutilisation des données fournies par le Groupe Randstad et Pôle emploi. L’équipe de Rennes 2 a dans un premier temps, exploré l’ensemble des données fournies sur l’emploi disponible et les quartiers et construit une maquette d’un outil permettant de simuler les différentes manières d’accéder d’un QPV à un emploi donné. La première proposition reposait sur la création d’une carte dynamique permettant de sélectionner des offres d’emploi à l’intérieur d’isochrones. C’est-à-dire des tracés représentant les zones accessibles en un temps donné selon le mode de transport disponible. Le prototype livré par les étudiant.e.s est un véritable site web à l’ergonomie intuitive et conviviale comprenant un outil de géovisualisation associé à plusieurs outils de visualisation des données. Quelques enseignements« Passer d’une expérimentation comme celle-ci à une utilisation efficace et généralisée des données au service des acteurs de la politique de la ville ne se fera pas si facilement. L’expérience des métropoles pionnières en matière de gestion des données (pensons à Rennes et Nantes par exemple) peut nous être utile. Des nombreux « prototypes » restent parfois dans les cartons. Il faut réunir des conditions bien particulières pour installer dans la durée un pilotage avec des données.Il faudra d’abord s’assurer que les données nécessaires qui ont été mises à disposition de façon expérimentale pourront l’être de façon pérenne. Cela peut prendre du temps. Et il y aura un coût. Ceci vaut aussi bien pour des acteurs publics (pensons par exemple aux extractions qui ont été faites par la CNAM) que des acteurs privés ...Il faudra aussi que des acteurs soient formés et puissent les utiliser. L’étape d’acculturation, si importante dans le programme « Data & Quartiers » devra être organisée partout, auprès des élus, des agents territoriaux ou des acteurs associatifs. Et devra parfois être répétée.Il faudra construire des outils techniques pour que la data science soit rendue accessibles aux acteurs de terrain : des interfaces, des flux et des tableaux de bord automatisés, des cartes interactives, des outils d’analyse « clef en main »… Tout ceci a un coût. L’intégration de ces coûts dans les prochains contrats de ville semble une nécessité.Il faudra aussi prendre en compte des enjeux démocratiques et politiques. Une collectivité territoriale, un ministère, un centre de ressources… ce n’est ni Google, ni Facebook, ni Amazon ! L’utilisation massive des données mise au service de l’intérêt général se doit de respecter des règles qui vont bien au-delà du simple respect de la loi, notamment en matière de protection des données. Des acteurs engagés dans des politiques qui concernent des quartiers et des publics sensibles doivent se doter d’une éthique de la donnée. »

• Data

• Inclusion

• Citizenship

• Innovation

Published on 28/10/22



Towards improved accessibility of INSEE data

Parliamentary workOn the occasion of an information report on the French National Institute for Statistics and Economic Studies (INSEE), the French National Assembly's Finance Committee takes stock of INSEE's open data policy.In application of the Digital Republic Act of 2013, " from now on, all INSEE information and data may be reused freely and without charge, including for commercial purposes... The provision of results and databases is systematically accompanied by all the metadata, including documentation, necessary for their proper use, in contrast to the practice organized by some private companies not necessarily delivering their metadata."The Insee website has recorded 30 million visits in 2022 (up 16% on 2021)."Through focus groups and user surveys, the website is being transformed, notably with the development of "datavisualization" and improved accessibility and readability of statistics (improved information location and navigation, full html versions available for all publications in addition to pdf versions, data fully accessible for download)".The parliamentary information report points out that "the open data policy has resulted in a net loss of revenue for INSEE, which has been fully compensated by the State".While INSEE continues to add to its data offering by opening APIs (programming interfaces), the Finance Commission's report regrets that some APIs are restrictive. It mentions the expectations of the research community and civil society players, such as theOuvre-boîte association, who are calling for a " more complete opening up of open data ". Insee data catalog in beta versionIn order to promote access to its open data offering, INSEE has developed a new data catalog, providing a variety of search and consultation functions. This catalog is currently available in beta version."The data published in the catalog are not updated as frequently as those on the INSEE website, and in the event of any discrepancies in content, the INSEE website will be the reference". At this stage, the stabilized version of the catalog is scheduled for release in mid-2024.

• Data

• Citizenship

Published on 19/02/24