Informed Decision-Making and Design: Big Data Applications from the Classroom to the Smart City
As ever-increasing amounts of data are available to the public, the ability to analyze and act on such data is a specialized skill that is becoming progressively important for designers and planners to master.
Since the 2012 U.S. Big Data Research and Development Initiative was launched by the Obama administration, much of this "big data" is open-source and available to anyone with an interest in accessing all kinds of information. The term "big data" may apply to a broad range of topics but is generally accepted as referring to data with more than 50,000 observations, larger than a Microsoft Excel spreadsheet can effectively handle.
Although so much data is accessible to the public, data analytics literacy is necessary to make sense of it, and make use of it, said Kieran Donaghy, regional science professor in CRP. Data analytics is the fastest-growing academic field at Cornell University: With a new Department of Statistics and Data Science in the School of Operations Research and Information Engineering (ORIE) in Ithaca, and a two-year Master of Science in Information Systems program at Cornell Tech, Cornell is preparing its students at all levels to stand at the forefront of understanding and implementing big data in a variety of fields — including within AAP.
In concert with Cornell Tech, AAP is developing a New York City–based specialization in urban technology, set to begin in fall 2020. The broader AAP community is using large swathes of data to contextualize research and discern underlying relationships that may aid in art or design — from data mapping and visualization to responding to climate change in northern European cities, and improving the lives of renters at risk for displacement in New York City.
Quantifying Urban Quality of Life Concerns
Emily Goldman (M.A. HPP '07, Ph.D. CRP '16) is the director of the Civic Innovation Fellowship, a program of BetaNYC, which seeks to empower communities to more actively participate in governance and decision-making using open data and civic technology.
"The main way we work with big data is that we try to break it down for the people who can make real use of it. One of our first target audiences was community boards," she said. Each community board — the smallest government agencies in New York City — is primarily composed of volunteers who help make decisions for their neighborhood and act as advisors to city agencies.
Goldman and her team created multiple tools to help community boards use data to support their recommendations, one of which is BoardStat. BoardStat collects 311 service requests (311 provides access to nonemergency city services and information about city government programs) and breaks the data down by community board, highlighting trends, hot spots, problematic buildings, top ten complaint types, and more.
Tenants Map, which scrapes and visualizes rent-stabilized data, enables the public to see the rent-stabilized building stock, and to visualize housing-related issues over that increasingly vulnerable stock. Mapping heat and hot water service requests over rent-stabilized housing, she said, can give an indication of how many rent-stabilized tenants are vulnerable to displacement.
Goldman said: "While community boards may hear from residents anecdotally what's going on in neighborhoods, BoardStat helps them quantify those concerns. They can pull specific information to show the extent and history of a quality of life concern. They can elevate that issue to the council district and use the information to strengthen funding requests in their annual Statement of District Needs."
Improving Workplace Processes
Recent graduate Andrés Gutiérrez (B.Arch. '15, M.S. '19) used his degrees in architecture and computer graphics to start a company, Comake, focused on improving processes for the modern workplace.
"Corporate data is doubling every 14 months, and roughly 80 percent of enterprise workflows are going to be cloud-based in 2020. Put those things together, and it's easy to understand why we increasingly struggle with information overload and have difficulty accessing the right info at the right time," Gutiérrez said.
Gutiérrez and fellow Cornellian Adler Faulkner '18 first received support for the idea that would eventually become Comake through Cornell's eLab accelerator program. After receiving a grant from the National Science Foundation, the pair was able to launch its cloud-based operating system, which consolidates web apps and accounts, as well as the work information within them, including files, messages, tasks, and contacts. The software automatically maps, interrelates, and contextualizes these components of work to create a collective knowledge graph. Comake automatically maps all this relevant information across apps to a given project or person.
Recognizing that patterns and similarities across projects, people, and teams can lead to valuable insights, Comake gives workplaces the opportunity to analyze these data to understand how people are working, to more easily repurpose knowledge, and to optimize outcomes.
"It's not just about establishing and understanding relationships between pieces of information, but also about using those relationships to help increase value, creativity, and innovation," Gutiérrez said.
Critically Examining Data in Context
A recently published book by Yanni Loukissas (B.Arch. '99), a former lecturer in the architecture program, explores the relationship between data and place — what Loukissas calls "locality."
All Data Are Local: Thinking Critically in a Data-Driven Society (MIT Press, 2019) emphasizes that data are made in specific conditions often for very specific audiences. Loukissas argues that it is a mistake to think that data can be universally understood without accounting for the contingencies of their creation and the inevitability of bias.
"The overriding lesson of the book," he said, "is to stop thinking about data as data sets. Set implies something complete and internally consistent, which can be transferred anywhere. Instead, we should start thinking about data settings, the places in which data are made, and the conditions in which they are used. We're in this new era where the dark side of data is revealing itself."
In one example, Loukissas attempted to create a data visualization of records from the New York Public Library. While trying to organize all their records of books, images, and newspapers, Loukissas discovered that the process for cataloging dates varied enormously over the course of the collection's history. In fact, he found more than 1,700 distinct date formats. "Maybe they only know the century, or it is in Roman numerals, or it includes the printer's name. This was an early indicator of the heterogeneity of data — rather than a big pile of info, each data point retained cultural artifacts from different practices that we need to learn about and understand."
Responding to Extreme Weather Events
Making the transition from regional scientist to data scientist was a natural step for Scarlett Zuo (Ph.D. RS '14), who first began collecting raw data on environment economics and learned to parse it using MATLAB. After graduation, she moved to Switzerland and joined CelsiusPro, a Swiss insurtech company that specializes in indexed insurance solutions to mitigate the effects of adverse weather, climate change, and natural catastrophes. Using a combination of data analytic tools, she orchestrated terabytes of weather data from various sources.
"The business problem I was trying to solve has two parts. First, farmers, businesses, and other consumers need insurance to protect them from big weather events and natural disasters. Second, insurance companies need to leverage weather data to set prices," Zuo said. She helped create a set of products that consume weather data and determine if and how much payout is required, allowing for customized parameters, such as what constituted a day with abnormally high temperatures. Using a drought insurance product as an example, once the parameters of a "heat day" and number of total heat days indicating a drought are set, the product can determine the likelihood of a drought in the insured period, and provide a corresponding insurance payout.
Zuo returned to the U.S. — and to Cornell — and is now the lead data engineer for Cornell Research Administration Information Services (RAIS). RAIS manages the IT systems that oversee research compliance and administrative systems at Cornell, such as workflow systems for proposal development, contract negotiation, and research protocol review. Working to consolidate multiple legacy systems into a streamlined single system, Zuo is helping to capture metrics data and integrate compliance and training to give faculty a better proposal development and project management experience.
Ethically Planning Smart Cities
With big data comes big responsibility. While urban planning classes are investigating ways to incorporate emerging methodologies into practice, Associate Professor Jennifer Minner also examines the potentially problematic side to data-driven planning, such as using GIS to predict or respond to abandonment. The algorithms these systems use are imperfect — incorporating graffiti, she said, is not a neutral way of detecting where there's so-called blight, and can actually create the conditions for contemporary redlining.
"There are a lot of ethical questions to the use of data," she said. "I first started teaching a class called Cities, Place, Technology to support students thinking expansively and critically about tech," helping them to think responsibly about the ethical application of data techniques such as machine learning in city planning.
Minner's area of research combines urban planning with historic preservation and tech. In her class Community Shaping Technologies, she focuses on how technology shapes communities and how communities can have agency in shaping and applying technology.
"What I want new planners and preservationists to do is think about how open data is, how is it used in the public interest, and how we ensure that we're helping with the equitable preservation and planning of communities."
Ryan Thomas, a third-year CRP Ph.D. candidate, also researches the role of technology in decision making and planning.
He is a teaching assistant for Stephan Schmidt's Advanced Topics in GIS class, where students begin to use large data sets and application programming interfaces.
Many students, Thomas included, are interested in planning to advance sustainability. Because of this factor, the use of scientific data sets is becoming more of an everyday skillset as future planners will need to use environmental models to decide how to locate urban developments in areas that are safe from hazards like sea-level rise or wildfires.
"There's a growing call for incorporating big data into the curriculum," Thomas said, adding that there are two responses: overwhelming excitement and major skepticism. "I think that both responses are valid and important. Planning and big data practitioners over the next several years will be grappling with the need to account for different levels of access to technology, computer literacy, issues of surveillance, and privacy while embracing new technologies."
Considering the Artificial Intelligence of Architecture
Assistant Professor of the Practice Martin Miller, architecture, is exploring the ideas of big data and artificial intelligence (AI) in his third-semester graduate studio, exploring artificial physic-based simulations, finite element analysis, evolutionary solvers, and neural networks.
The students are tasked with analyzing data from an entire city rather than a particular site, working as a collective to make sense of the vast amounts of information available on everything from food networks to police reports and transportation data. In spring 2019, they created a collection of 140 maps including Citi Bike minute-on-minute animations. In the fall, his studio shifted focus to San Francisco and began to overlap the data sets to see what sort of patterns emerge and use these patterns to determine need and circumstance for architectural intervention.
"While many may think that the AI tools will take over and run design and decisions will be decided by algorithms, this is just not true," Miller said. "I can delegate that decisionmaking power to a computer and give it my opinion on particular conditions and how to react when facing those millions of turning points."
Meeting Environmental Demands on the Building and Urban Scale
Assistant Professor Timur Dogan directs the Environmental Systems Lab, where he creates software that combines physical data — such as how much sunlight a building receives and how warm the building is as a result — and behavioral data — how people interface with the building and when did it become hot enough that they opened a window — to simulate a building's energy use. These research outcomes aim to empower architects and urban designers to develop more sustainable and livable design proposals.
Dogan's latest work in evidence-based design is creating software to model the attractiveness and comfort of outdoor spaces in cities, which may become increasingly relevant in a warming climate. Studies show that the use of outdoor space is correlated with outdoor comfort, so Dogan uses a new outdoor comfort metric called the Universal Thermal Climate Index while examining city block data from New York City.
Studying data from Google Places, Dogan can model the walkability of a neighborhood and alter certain design parameters to add buildings, footpaths, parks, and cafés, or even reroute streets to simulate what space utilization could look like and determine how to meet the city's demands. "If you plan with outdoor comfort in mind, biking and walking become much more feasible as a mode of transportation," he said.
Mapping Technology in Visual Art
Mapping projects that incorporate data is a specialty of transdisciplinary artist Jaret Vadera, assistant professor of the practice in the Department of Art.
"I think a lot about how meaning is generated through process and how tech influences perception, and I'm very interested in the relationship between noise and signal. There is a fine line between abstraction and representation, and this applies to a number of different technologies," Vadera said.
Much of Vadera's work deals with data, translation, power, and the ways in which algorithms and imaging systems shape and control perceptions. One recent installation, This That and the Third, integrated visual aesthetics of infographics and Rorschach tests, produced in black vinyl on a white wall.
"For each one of those forms, I started with a sentence or phrase, often a mistranslation, an idea that's harder to depict or represent in a finite way. I do a search using a search engine, find images that correspond with each one of those words, subtract color, and amalgamate them," he said. The end result is a combination of organic and technological forms. Vadera then plots the image's server origin on an invisible world map that ties into each one of the forms.
In his classes, he said: "The idea of mapping often feels, or is perceived as, very static. But using mapping is a way of thinking about translation, understanding how translation works, and giving visual literacy to the students in a different way."
Research at the Intersection of Architecture and Science
For nearly 15 years, Associate Professor Jenny Sabin's research has focused on data-driven design, visualization, and simulation across disciplinary boundaries. She continues to investigate the intersection of architecture and science, both on and off-campus.
As the current designer-in-residence for Microsoft Research, she and her team at Jenny Sabin Studio worked on Ada, a year-long project with researchers to open the first built structure driven in real-time by artificial intelligence.
"It draws from a massive database of live sentiment data," Sabin said. "There's a network of cameras sensing people's reactions, which in turn reflect individual and collective sentiment throughout the building."
Sabin, who is the associate dean for design initiatives, says that discussing data — how we're working with it, visualizing it — has been essential to her individual work and collaborations.
"It's a bridge for me and my teams in my lab at Cornell and at my practice to collaborate with material scientists, engineers, and biologists at an important meeting point. Big data is completely changing how science is done. Science is no longer solely about repeating discrete events to target particular results. Instead, we are sifting and searching for trends and behavior through data," she said. "We've been successful in applying our design skills to the scientific process. It's an amazing point of connection."
By Jennifer Wholey