A New Map of the Milan Urban Region Through Mobile Phone Data

Abstract

Interpretative tools for identifying mobility practices in the contemporary cities are needed, not only for the limitations of conventional data sources1, but also because new forms of mobility are emerging, describing new city dynamics and timevariations in the use of urban spaces by temporary populations.These mobility practices result from the combination of physical and virtual mobility, leading to new, mixed forms of daily, residential, and travel mobility (Flamm & Kaufmann 2006).The transformations to practices of mobility question the available sources and open up toward operative challenges: the analysis of the space-time variability of mobility practices, while offering a representation of various urban rhythms2 and identifying different mobile populations, remains difficult to achieve with traditional data sources. In this perspective, an interesting contribution may be provided by mobile phone network data as a tool for real-time monitoring of urban dynamics and mobile practices. In recent years, several research projects focused on the potentiality of mobile phone traffic data as promising sources for the analysis, visualization and interpretation of people’s presence and movements in urban spaces. The contribution that may come from mobile phone network data for the analysis and description of urban practices, seems of great interest, due to its fine spatial and temporal resolution.As tested in several studies (Ratti et al. 2006; Ahas & Mark 2005; Soto & Frías-Martínez 2011a,b; Reades et al. 2007; Gonzalez et al. 2008), passive monitoring of anonymous telephone traffic represents a valuable alternative to traditional methods, because it can simultaneously overcome the limitations of the detection latency time typical of traditional data sources and take advantage of the ubiquity of mobile phone networks and the pervasive diffusion of mobile devices.3 Among different survey methodologies4, the researches that focused on the analysis of aggregated mobile phone data are characterized by two different profiles and purposes: mapping mobile phone activity in urban contexts (Ratti et al. 2006; Sevtsuk & Ratti, 2010), and visualizing urban metabolism (Wolman, 1965; Acebillo & Martinelli, 2012; Brunner, 2007).The first approach, named Mobile landscape approach, focuses on the relationships between mobile phone measures and people’s daily activities in cities (Ratti et al., 2006; Sevtsuk & Ratti, 2010). The aim is to understand patterns of daily life in the city, using a variety of sensing systems (mobile phone traffic intensity, location-based data as GPS devices, wireless sensor network) and to illustrate and to confirm the significant differences in the distribution of urban activities at different hours, days and weeks. Graphic representations of the intensity of urban activities and their evolution through space and time, based on the geographical mapping of mobile phone usage at different times of the day (Ratti et al., 2006), are the main output of the Mobile Landscape approach.The approach based on handsets’movements studies the relationships between location coordinates of mobile phones and the social identification of the people carrying them (as Social Positioning Method and its possible applications in the organization and planning of public life proposed by Rein Ahas and Ülar Mark, 2005).In this framework an interesting issue regards the classification of urban spaces according to their users’ practices and behaviors in the use of cell phones (Soto & Frías-Martínez, 2011a).According to Soto and Frías-Martínez (2011a, 2011b) city areas are generally not characterized by just one specific use, and for this reason they introduce the use of c-means, a fuzzy unsupervised clustering technique for land use classification, which returns for each area a certain grade of membership to each class.Even if, from a technical point of view, both the aforementioned approaches are based on the analysis of aggregated data and traffic volume detected on towers of the network, the loss of the traces of the origins and destinations of individual movements does not appear relevant for estimate the distribution patterns of the population in different time slots considered for the survey.Using mobile phone data for monitoring urban practices, both approaches show that phone calls are closely related to population density in urban areas (Ratti et al. 2006; Sevtsuk & Ratti, 2010; Ahas & Mark, 2005; Reades et al., 2007), even if additional evidence is needed to specify how mobile network signals can be used to characterize and map different urban domains and their occupants and how this tool could support urban planning and policy.According to this background, our research focused on mapping and visualizing the changing city by means of these new sources, characterized by a high temporal and spatial resolution.KeywordsMobile Phone DataMilan UrbanMobility PracticesTraditional Data SourcesErlang DataThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.