I began the design process for this project by completing four explorations. Each exploration focused on a key objective that was important to investigate prior to proposing and designing final prototypes. The following documents my process and reflections for each exploration.
- Exploration 1: How to visualize converging variables
- Exploration 2: How to design for weighted preferences
- Exploration 3: How to visualize hierarchical scales of space
- Exploration 4: What interaction may contribute
DESIGN EXPLORATION 1
How to visualize converging variables
OBJECTIVE
One of the earliest challenges in this investigation was to determine a strategy for visualizing multiple variables in the same information space. Designers and cartographers have codified various methods for plotting information on maps throughout history. Jacques Bertin’s influential Semiology of Graphics (1983) provided a foundation for data, pattern, and information visualization in map form. I found that heat maps are good for expressing both location and intensity. However they typically plot a single variable on the map. Because my focus was to combine multiple variables, I started by collecting heat maps from a few different online sources and created small studies that combined their data using different graphic representations.
One of the earliest challenges in this investigation was to determine a strategy for visualizing multiple variables in the same information space. Designers and cartographers have codified various methods for plotting information on maps throughout history. Jacques Bertin’s influential Semiology of Graphics (1983) provided a foundation for data, pattern, and information visualization in map form. I found that heat maps are good for expressing both location and intensity. However they typically plot a single variable on the map. Because my focus was to combine multiple variables, I started by collecting heat maps from a few different online sources and created small studies that combined their data using different graphic representations.
REFLECTIONS
After completing these studies, I found that they fell into three loosely-defined categories: 1) those that utilize symbols and grid structures, 2) those that utilize convergence of surface area, and 3) those that expressed convergence through area contours. I found there to be trade-offs between the various methods of visualizing data combinations. Meng, Zipf, and Winter (2008) note that the choice to use mimetic representation (realistic, representative of that which it symbolizes, such as landmarks on tourist maps) allows for much quicker recall and comprehension of correspondence to the real environment. The use of random symbols, on the other hand, are interchangeable between any of the variables, though the use of a key is required to decipher the identity of each variable.
Geographers have written about such difficulties in visualization as well. MacEachren (1992) notes that the type of visualization chosen should respond to the data or phenomena being represented. The “...continuum relates to character of variation in the phenomenon across space. Some phenomena (e.g., tax rates) can vary quite abruptly as political boundaries are crossed while others (e.g., gallons of ground water pumped for irrigation per county) can exhibit a relatively smooth variation quite independent of the units to which data are aggregated,” (1992, p. 7).
After completing these studies, I found that they fell into three loosely-defined categories: 1) those that utilize symbols and grid structures, 2) those that utilize convergence of surface area, and 3) those that expressed convergence through area contours. I found there to be trade-offs between the various methods of visualizing data combinations. Meng, Zipf, and Winter (2008) note that the choice to use mimetic representation (realistic, representative of that which it symbolizes, such as landmarks on tourist maps) allows for much quicker recall and comprehension of correspondence to the real environment. The use of random symbols, on the other hand, are interchangeable between any of the variables, though the use of a key is required to decipher the identity of each variable.
Geographers have written about such difficulties in visualization as well. MacEachren (1992) notes that the type of visualization chosen should respond to the data or phenomena being represented. The “...continuum relates to character of variation in the phenomenon across space. Some phenomena (e.g., tax rates) can vary quite abruptly as political boundaries are crossed while others (e.g., gallons of ground water pumped for irrigation per county) can exhibit a relatively smooth variation quite independent of the units to which data are aggregated,” (1992, p. 7).
DESIGN EXPLORATION 2
How to design for weighted preferences
OBJECTIVE
A second series of graphic investigations focused on creating formal studies that represented relative weight among chosen preferences. Taking the fixed-total principal from Thomas Saaty’s Analytical Hierarchy Process, these studies explored ways in which weighted preferences can be visualized in terms of parts- to-whole (much like a pie-chart).
Below are sample visualizations that investigate the concept of prioritized variables in terms of prominence. In each study, an increase in the importance or weight of one variable results in the decrease in importance of all others.
A second series of graphic investigations focused on creating formal studies that represented relative weight among chosen preferences. Taking the fixed-total principal from Thomas Saaty’s Analytical Hierarchy Process, these studies explored ways in which weighted preferences can be visualized in terms of parts- to-whole (much like a pie-chart).
Below are sample visualizations that investigate the concept of prioritized variables in terms of prominence. In each study, an increase in the importance or weight of one variable results in the decrease in importance of all others.
REFLECTIONS
These studies all utilized visualization methods distinct from the cartographic studies in the previous exploration. Designing a user interface that allows a user to re-assess his or her preferences is crucial to the user’s understanding of the relationship between search criteria and the resulting visualization of desirable places on the map. This re-assessment might result from changing the variables displayed on the map, personal experience through visiting the location in-person, or recommendations from trusted friends.
Along with visualizing the preferences, the system might also help a user further define his or her preferences. Achieved through recommendation prompts, such solutions might use contextual search cues that offer the user variations on search terms and allow the user to either confirm or revise his or her initial search criteria.
These studies all utilized visualization methods distinct from the cartographic studies in the previous exploration. Designing a user interface that allows a user to re-assess his or her preferences is crucial to the user’s understanding of the relationship between search criteria and the resulting visualization of desirable places on the map. This re-assessment might result from changing the variables displayed on the map, personal experience through visiting the location in-person, or recommendations from trusted friends.
Along with visualizing the preferences, the system might also help a user further define his or her preferences. Achieved through recommendation prompts, such solutions might use contextual search cues that offer the user variations on search terms and allow the user to either confirm or revise his or her initial search criteria.
DESIGN EXPLORATION 3
How to visualize hierarchical scales of space
OBJECTIVE
For this exploration, I anticipated the visual relationships between the macro-, city-wide view and the micro-, neighborhood-specific view. These relationships between multiple scales are complex because certain information is appropriate at each scale, yet the viewer still needs to understand that the two representations are related. The city-wide view typically involves generalized visualizations of both place and human activity, while the neighborhood-specific view needs to show more specific data and a concrete depiction of place. I expect this higher level of fidelity includes aerial or satellite photos, as well as user-generated images and content.
For this exploration, I anticipated the visual relationships between the macro-, city-wide view and the micro-, neighborhood-specific view. These relationships between multiple scales are complex because certain information is appropriate at each scale, yet the viewer still needs to understand that the two representations are related. The city-wide view typically involves generalized visualizations of both place and human activity, while the neighborhood-specific view needs to show more specific data and a concrete depiction of place. I expect this higher level of fidelity includes aerial or satellite photos, as well as user-generated images and content.
REFLECTIONS
The decision to keep the visualization more generalized at a macro-view is deliberate. This strategy prevents the user from relying too heavily on the chosen representation method at the chosen scale and is analogous to MacEachren’s description of “contour crispness,” as a means for depicting certain and uncertain geographic boundaries. He uses the US–Canada border as a certain boundary, whereas the 1990’s Kuwait–Iraq border as an uncertain boundary (MacEachren, 1992, p. 6).
Through discussions with my committee as well as peers, it became clear that a user would approach these multiple scales of representation with different cognitive perspectives. The macro-view, with its presentation of general patterns and converging variables over time, encourages an analytical mindset. The neighborhood-specific view, with its high-fidelity data of satellite images and textual content elicits more emotional responses from a user. Both are highly useful in making decisions, so I chose to carry these aspects into the final prototype designs.
The decision to keep the visualization more generalized at a macro-view is deliberate. This strategy prevents the user from relying too heavily on the chosen representation method at the chosen scale and is analogous to MacEachren’s description of “contour crispness,” as a means for depicting certain and uncertain geographic boundaries. He uses the US–Canada border as a certain boundary, whereas the 1990’s Kuwait–Iraq border as an uncertain boundary (MacEachren, 1992, p. 6).
Through discussions with my committee as well as peers, it became clear that a user would approach these multiple scales of representation with different cognitive perspectives. The macro-view, with its presentation of general patterns and converging variables over time, encourages an analytical mindset. The neighborhood-specific view, with its high-fidelity data of satellite images and textual content elicits more emotional responses from a user. Both are highly useful in making decisions, so I chose to carry these aspects into the final prototype designs.
DESIGN EXPLORATION 4
What interaction may contribute
OBJECTIVE
For my final set of explorations, I wanted to see ways in which combinations of digital behaviors and interactions could facilitate both an understanding of the abstract representations of data as well as the relationship between multiple scales at which the system represents information. I constructed small, interactive online prototypes using Proto.io, and tested them on an iPad to get a sense of the actual size and setting in which the tool would be used.
For my final set of explorations, I wanted to see ways in which combinations of digital behaviors and interactions could facilitate both an understanding of the abstract representations of data as well as the relationship between multiple scales at which the system represents information. I constructed small, interactive online prototypes using Proto.io, and tested them on an iPad to get a sense of the actual size and setting in which the tool would be used.
REFLECTIONS
A clear benefit of interaction with the visualizations is that it gives the user control over the pace of change and relative focus for how the visualizations are shown. Additionally, the ability to zoom in and out of scale levels as a means of transition increases the chance that a user will connect the two levels of cartographic abstraction.
Finally, providing easy opportunity for the user to adjust his or her preferences through interaction (and in turn receive immediate feedback in the form of map visualizations) is helpful in understanding the relationship between the combination of what he/she wants in a neighborhood environment and what is actually available.
A clear benefit of interaction with the visualizations is that it gives the user control over the pace of change and relative focus for how the visualizations are shown. Additionally, the ability to zoom in and out of scale levels as a means of transition increases the chance that a user will connect the two levels of cartographic abstraction.
Finally, providing easy opportunity for the user to adjust his or her preferences through interaction (and in turn receive immediate feedback in the form of map visualizations) is helpful in understanding the relationship between the combination of what he/she wants in a neighborhood environment and what is actually available.
See also: