Visual(izing) Data

 

In the fall of 2014, Contour released a call for visual presentations of data among PhD students of the EPFL and international participants of the Laboratory Basel’s 2014 colloquium Intercultural Dimensions of Territory. The call was also open to interested contributors from other institutions. After receiving 17 submissions, we are pleased to release these images to the public, with the original short text that accompanied them. Following reception and acceptance of the images, we decided to ask the authors to reflect upon the following questions:

  1. How does this visualization help you to understand and analyze your data?
  2. Were you in any way surprised by what was revealed about your data through its visualization?
  3. How does this visualization help you to communicate your findings from your data and who specifically is the targeted audience?
  4. What is the role of color in visually presenting your data? Would any color have sufficed?

Over the course of the next couple months, the Contour web site will present the visualizations along with their accompanying short texts and the authors’ responses to our questions as images of the week (in the right column on the main page of the site) and indexed/archived on the Contour Tumblr site. Essays written by several of the journal editors and inspired by the contributions will be posted to the site as they become available.

Essays
Doyle, Michael. What do we touch when we visualize data?

The organizing editors of this call would like to thank each of the contributors to the Visual(izing) Data call.
Andrew Becker, Processor Architecture Laboratory, EPFL, Switzerland
Iva Bojic, MIT, USA
Nancy Couling, Laboratoire Basel, EPFL, Switzerland
Antonin Danalet, Transportation and Mobility Laboratory, EPFL, Switzerland
Michael Doyle, Laboratory for Environmental and Urban Economics, EPFL, Switzerland
Andrea Galli, Carlo Ratti Associati, Italy
Elnaz Ghazi, University of La Sapienza, Italy
Mary Katherine Heinrich, Institute for Advanced Architecture of Catalonia, Spain
Michael H. Herzog, Brain Mind Institute, EPFL, Switzerland
Darius Karacsony, Laboratoire ALICE, EPFL, Switzerland
Marc M. Lauffs, Brain Mind Institute, Psychophysics Laboratory, EPFL, Switzerland
Michael H. Herzog, Brain Mind Institute, EPFL, Switzerland
Marlène Leroux, EPFL, Switzerland
Lorenzo Massimiano, University G.d’Annunzio of Chieta-Pescara, Italy
André Ourednik, Chôros Laboratory, EPFL, Switzerland
Silvia Paldino, University of Calabria, Italy
Trevor Patt, Media and Design Laboratory, EPFL, Switzerland
Love Råman Vinnå, Physics of Aquatic Systems Laboratory, EPFL, Switzerland
Siobhan Rockcastle, Interdisciplinary Laboratory of Performance-Integrated Design, EPFL, Switzerland
Dario Rodighiero, Digital Humanities Laboratory / Chôros Laboratory, EPFL, Switzerland
Selena Savic, Media and Design Laboratory, EPFL, Switzerland
Mandana Sarey Khanie, Interdisciplinary Laboratory of Performance-Integrated Design, EPFL, Switzerland
Stanislav Sobolevsky, MIT, USA
Isabel Vollenweider, IRP Chair on Spinal Cord Repair, EPFL, Switzerland

Contour recommends using the following convention for citing these images:
Paldino, Silvia et al. (2015). Circos. In Visual(izing) Data. Retrieved from http://contour.epfl.ch/visualizing-data/ on March 15th 2015.


Paldino-et-al_Circos2015

Circos

Silvia Paldino, Stanislav Sobolevsky, & Iva Bojic (University of Calabria, Italy & MIT, USA)

(Also appeared in: Paldino, Silvia et al. (2015). Urban Magenitism through the Lens of Geotagged Photography. arxiv preprint arxiv: 1503.05502, made available here by permission of the first author)

“Circos” is in fact a software package for visualizing data in a circular layout, which emphasizes the links between the different components of the diagram. Although it was originally designed for visualizing genomic data, by now it can be actually used for plotting every kind of data that describe relationships. In our case, we used this representation to diagram an origin-destination network between different cities.

Contour: How does this visualization help you to understand and analyze your data?
PS&B: This visualization helps us to see immediately the mobility flows between our considered cities in an intuitive way. A regular data table would be much more difficult to interpret.

Contour: Were you in any way surprised by what was revealed about your data through its visualization?
PS&B: It was quite surprising to find links between American and European cities to be surprisingly asymmetric. Namely, links going from American origins to EU destinations are on average stronger than the ones going in the opposite direction.

Contour: How does this visualization help you to communicate your findings from your data and who specifically is the targeted audience?
PS&B: The visualization helps us to communicate our findings in an intuitive way as it is much easier to visually capture relative strength and the patterns among the 45 flows between each pair of 10 cities compared to a lengthy numerical table. With this visualization we can immediately understand what cities are more linked to each other and what cities have a weaker connection. The targeted audience is the scientific community, in particular big data, complexity and network science researchers, urban planners and touristic sector stakeholder as well as for the general public.

Contour: What is the role of color in visually presenting your data? Would any color have sufficed?
PS&B: The color here is important to differentiate flows related to each city from the others. Although in general any set of different colors could have sufficed, it is quite important to have the color palette smooth and distinctive.


ATLAS_with_legend3

Atlas

Antonin Danalet (Transportation and Mobility Laboratory, EPFL, Switzerland)

(Also appeared in: A. Danalet, B. Farooq and M. Bierlaire. A Bayesian Approach to Detect Pedestrian Destination-Sequences from WiFi Signatures, in Transportation Research Part C: Emerging Technologies, vol. 44, p. 146 – 170, 2014., made available here by permission of the first author)

La carte permet de mieux penser le piéton, de l’étudier et le comprendre. Les cartes pour la voiture sont nombreuses. Elles véhiculent une image de liberté, de voyage, de vacances. Le piéton dans ses chemins, ses habitudes, ses comportements et ses lignes de désir est moins cartographié parce qu’il est difficile à suivre, imprévisible, indétectable. La carte est ce qui permet de l’étudier, de le mettre dans des cases un peu rigides, de faire ressortir certains traits, même réducteurs, en partie faux peut-être, insuffisants pour comprendre l’ensemble de son comportement, mais des traits qui permettent de l’appréhender, de le structurer, de le penser.

Contour: How does this visualization help you to understand and analyze your data?
AD: In order to understand pedestrian activity choice, we collected WiFi traces on campus. GPS cannot be used indoors and 789 WiFi access points are already installed on campus. Unfortunately, WiFi localization lacks precision and it has to be merged with other data sources in order to detect activity episodes. Map data shown in this picture are particularly useful to compute the shortest path between two points of interest and to compute the travel time between these two points.

Contour: Were you in any way surprised by what was revealed about your data through its visualization?
AD: The EPFL pedestrian network is particularly dense, with its different floors. The structure of the network in the Rolex Learning Center is particularly interesting, showing no access by the side of the building and a very different structure of paths compared to the traditional grid.

Contour: How does this visualization help you to communicate your findings from your data and who specifically is the targeted audience?
AD: Studying pedestrian behavior is particularly challenging. People walk in a denser, multidirectional network with less rules than in a road network. Showing the density and complexity of a pedestrian network is a way to explain the value of such data, the complexity of the modeling and the short distances between different points of interest.

Contour: What is the role of color in visually presenting your data? Would any color have sufficed?
AD: Without color, the image only expresses the shape of EPFL pedestrian network. The color gives information on the shortest paths and the priority of the different segments of the paths. The more red it is, the more it is used.


Patt_Qinglonghuzhen

Qinglonghuzhen, Peri-Urban Beijing

Trevor Patt (Media and Design Laboratory, EPFL, Switzerland)

Alain Badiou writes that “the model is that which allows us to think through participation,” deliberately separating it from the task of empirical grounding. Working within a complex generative model requires visual feedback that enables the designer to interpret the quality of intermediate results. This model parses local behaviors that coordinate parcel density and adjacency guidelines for a large-scale residential development (~700,000 m2) over a sequence of iterative scenarios. Built into the model, are customized display pipelines for highlighting selected details as well as a protocol for collecting and displaying statistical trends across the iterations concurrently with the spatial design in the same window. Shown here is a histomap that illustrates the volatility of programmatic assignments over 75 iterations and reveals temporal trends that give better understanding of the inner workings of the model and how it compares to other permutations of the code.

1. Badiou, The Concept of Model, trans. Zachary Luke Fraser and Tzuchien Tho, (Melbourne: re.press, 2007),
http://re-press.org/books/the-concept-of-model/. p92
2. Bogost, Ian, Persuasive Games: The Expressive Power of Videogames (Cambridge: MIT Press, 2007). p63-64

Contour: How does this visualization help you to understand and analyze your data?
TP: The image shows one frame of a dynamic masterplanning process with a number of localized rules and behaviors. The complexity of interactions and the number of elements involved make it challenging to connect the process to the result. A visualization that presents the history of state changes at once and alongside the changing states helps identify emergent trends or patterns that are impossible to spot from the process alone.

Contour: Were you in any way surprised by what was revealed about your data through its visualization?
TP: The visualization was used as a design tool to adjust parameters or test the efficacy of the generative behaviors. After many iterations it no longer has many surprises, but often I would be surprised to find that certain relationships persisted even when parameters were varied widely; behaviors that I thought were central to the process turned out to be less essential than predicted.

Contour: How does this visualization help you to communicate your findings from your data and who specifically is the targeted audience?
TP: Because the individual behaviors have such complex interaction, a simple listing of them does little to actually describe the procedural model, which is better explained through effects—transitions and tendencies. This visualization draws attention to these temporal patterns and also makes them easier to understand and identify. (Of course, it’s also helpful in cases where video media is not available).

Contour: What is the role of color in visually presenting your data? Would any color have sufficed?
TP: The color themselves don’t matter, though the relationships between them do. In this image, color is used to represent a use condition that might change. I wanted to create groupings that differentiated built volumes (blues) and open space (greens) but not to create too sharp a division between categories in order to maintain the uncertain and changeable potential of each element.


Cumulative Contrast. Rockcastle, 2015

Cumulative Contrast

Siobhan Rockcastle (Interdisciplinary Laboratory of Performance-Integrated Design, EPFL, Switzerland)

Daylight is essential to our visual interpretation of architecture. Characteristics such as brightness and contrast create a sense of spatial depth and material texture through the interaction between surface, void, and shadow. While these characteristics are fundamental to our evaluation of interior space, they are often regarded as purely qualitative and there is a lack of research into the development of quantitative or objective criteria to assess them. Furthermore, our visual perception is largely influenced by ephemeral conditions in the surrounding environment such as sky type, time-of-day, and time-of-year. This image represents the cumulative measure of contrast in a top lit space across a single day in June. The color scale, from blue to red shows where contrasting lighting levels occurred throughout the space and the dynamism of these visual effects as a result of ephemeral environmental conditions.

Contour: How does this visualization help you to understand and analyze your data?
SR: This visualization is the product of a dynamic contrast analysis across 56 annual daylight renderings under sunny sky conditions at 42°N latitude. It shows the cumulative strength of contrast between areas of light and dark resulting from sunlight penetration at the specified latitude and more specifically, where these accumulations occur within the selected view direction. It is intended to represent areas of the image which are visually interesting due to the complexity of daylight composition across space and over time.

Contour: Were you in any way surprised by what was revealed about your data through its visualization?
SR: What is interesting about this form of data representation is that it reveals spatial information about contrast over time, which is inherently difficult to predict as we cannot visualize the dynamics of sunlight projections as they move through space. With the exception of visual comfort metrics, which account for light which reaches the eye, most researchers look at daylight performance across a horizontal surface at a single temporal instance, or, if dynamically, then they search for performance in terms of uniformity. This method of visualization seeks to uncover the complexity of daylight composition over time. In this sense, every image we output is surprising, delightful, and reveals unpredicted information about the instability of ‘performance’ on a dynamic scale.

Contour: How does this visualization help you to communicate your findings from your data and who specifically is the targeted audience?
SR: We are particularly interested in using images to communicate the spatial implications of a dynamic contrast metric because the resulting data is intended to help architects understand where sunlight composition produces the strongest visual effects and how these effects change over time. When combined with a temporal map, this visualization methods is a powerful communication tool for an audience which is inherently visual in both training, sensibility, and practicality.

Contour: What is the role of color in visually presenting your data? Would any color have sufficed?
SR: This is a very interesting question and one that would spark an interesting debate between scientific and design disciplines. The color map used in this data visualization is a default ‘jet’ produced through Matlab. It uses a linear gradient from dark blue-cyan-green-orange-red and unlike monochromatic scales (i.e. black to white), it allows us to differentiate a broad range of values with high resolution. This means that values which fall in the middle of the spectrum can be a range of cyan to orange instead of just grey. The problem with this scale is that data is often misinterpreted due to psychological associations with temperature (blue as ‘cold’ and red as ‘hot) or regulation (green as ‘go’or ‘good’ and red as ‘stop’or ‘bad’). The authors are currently exploring the use of custom color scales to better communicate the intended implications of our data. What color represents visual interest?


3D Attractiveness Map of New York City. Paldino & Sobolevsky, 2015

3D Attractiveness Map of New York City

Silvia Paldino (University of Calabria, Italy)
Stanislas Sobolevsky (MIT, USA)

(Also appeared in: Paldino, Silvia et al. (2015). Urban Magnetism through the Lens of Geotagged Photography. arxiv preprint arxiv: 1503.05502, made available here by permission of the first author)

People take pictures in some particular places that they consider important for some reasons. Then, using geotagged pictures data from photosharing websites, it is possible to map the most photographed places (hotspots), redrawing maps through people perceptions, going beyond traditional boundaries. A 3d map shows also in what measure each hotspot in the city is more or less “important” than the others. The visualization tells the story of New York City through the lens of geotagged pictures taken by visitors: the most photographed place is Time Square, while the Bronx, that does not attract that many people taking pictures, is not present on the map, except for the Yankees stadium.

Contour: How does this visualization help you to understand and analyze your data?
P&S: This visualization helps to identify the most photographed places in New York City, their location and the amount of attention they get from tourists.

Contour: Were you in any way surprised by what was revealed about your data through its visualization?
P&S: It was quite surprizing to notice that Time Square is absolutely the most photographed place in New York City and Time Square activity has a big difference from other important places, like Statue of Liberty or the museums. All the hotspots are located in Manhattan, while the rest of the city is quite “dark”. Really interesting is the only one peack in the Bronx, representing the Yankees stadium.

Contour: How does this visualization help you to communicate your findings from your data and who specifically is the targeted audience?
P&S: The visualization helps us to communicate our findings in a very intuitive way. In fact, in a 3D map we can visualize more information: the coordinates (longitude and latitude) and the ranks of the places. The pictures related to the highest bars also help people to know what specific place correspond to those coordinates. The target audience is the scientific community, in particular big data, complexity and network science researchers, urban planners and touristic sector stakeholder as well as the general public.

Contour: What is the role of color in visually presenting your data? Would any color have sufficed?
P&S: The color here is important to differentiate the density of people’s activity in each place. Although in general any set of different colors could have sufficed, it is quite important to have the color palette smooth and distinctive. The colors are hotter when the density is higher. Colors can give a good perception of density that otherwise only the bar with one color cannot provide.


SO.MA. Massimiano, Galli & Ghazi, 2015

SO.MA.

Lorenzo Massimiano (University G. d’Annunzio of Chieti-Pescara, Italy)
Andrea Galli (Carlo Ratti Associati, Italy)
Elnaz Ghazi (University of La Sapienza, Italy)


Every human being is composed of a set of elements, some visible and other invisible, that contributes to define its specific identity.
Over the centuries, science has explored all these aspects, constantly overcoming the boundaries of what was previously considered invisible and unknown. Today, thanks to digital technology, we are able to visualize our “virtual impact”, the ability to influence the real world through the virtual world.
The project SO.MA., using the graph of connections among friends on your Facebook profile, wants to stimulate speculations about this new and yet unknown aspects of our contemporary identity.

This text is a summary of the description of SO.MA., one of the winning project of Reshape Digital Craft Competition 2014. For further information, please visit Yourshape.com.

Contour: How does this visualization help you to understand and analyze your data?
P&S: The visualization that we used in the SO.MA design project shows all the connections between my friends on Facebook, through an algorithm developed by the Gephi Platform. The possibility to view my network in this way allows me to grasp at a glance the relational dynamics that gravitate around me, around my “virtual identity”, giving me back unknown information on the relevance and the impact of my actions in the virtual world (and not only: in fact, they also influence the real one).

Contour: Were you in any way surprised by what was revealed about your data through its visualization?
P&S: Ordering graphics in sequence you can learn about the evolution of your “virtual identity”, through an image that grows and transforms itself as if it was alive. What surprised us the most was to notice the aesthetic similarities between these images and experiments on simple organisms growth, such as molds. The two phenomena in fact seem to have common characteristics. If in the course of history, thanks to the improvement of technological tools, we have been able to study phenomena both on small and large scale, understanding that there is a relationship between the micro and macro cosmo, today we begin to recognize similarities between the real and the virtual cosmos.

Contour: How does this visualization help you to communicate your findings from your data and who specifically is the targeted audience?
P&S: Through this visualization, any Facebook user can observe their own “virtual portrait”. However, especially companies that make an extensive use of the social networks may be particularly interested in having a graph that periodically shows them not only how much they have grown, but also in which area of the graph, so to to direct future interventions towards specific targets.

Contour: What is the role of color in visually presenting your data? Would any color have sufficed?
P&S: In some variants of the model, the data are presented in different colours depending on the self organization of the network in clusters that the most of times correspond whith categories in which we usually divide friends in our Facebook profile: close friends, relatives, acquaintances, classmates, etc. This allows us to have more detailed information on the structure of the network, being able to find correlations in a simple and immediate way.