The Civic Data Science NSF REU Site began in the summer of 2017 when it was run in conjunction with the Data Science for Social Good program in Atlanta. Project descriptions from CDS and DSSG from 2014, 2015, 2016, and 2017 can be found here. Additional details about the project are archived in the blog.
Food for Thought: Analyzing Public Opinion on the Supplemental Nutrition Assistance Program Atlanta Community Food Bank
Housing Justice Atlanta Legal Aid Society
Cycle Atlanta: Seeing Like a Bike Atlanta Regional Commission
Building Energy Analytics Georgia Tech Facilities Management
UN Climate Challenge: Predicting & Alleviating Road Flooding in Senegal UN Global Pulse
Food for Thought: Analyzing Public Opinion on the Supplemental Nutrition Assistance Program
This Data Science for Social Good team is working in conjunction with the Atlanta Community Food Bank to create a project that analyzes patterns in food controversies and messaging in order to support more robust public communication. The DDSG + ACFB project is focused on analyzing the coverage of SNAP on a national and state level, summarizing key themes across media coverage, and providing a comprehensive data driven assessment of the most influential and effective messages about SNAP. Students will work together to build data sets, analyze initial data sets, and develop effective tools and techniques that can be used by the Atlanta Community Food Bank moving forward to find and acquire data.
The Atlanta Legal Aid Society provides free civil legal services to low-income people to improve their “social, political, and economic conditions.” Since their founding in 1924, they have won many landmark cases and coined the term “predatory lending.” Atlanta Legal Aid is currently involved in a housing justice lawsuit against Harbour Portfolio Advisors, a company that profits from selling homes through “contract-for-deed” agreements. These land contract arrangements are frequently marketed as an alternative way for low-income people to buy homes if they may not qualify for a traditional mortgage. They typically involve a down payment, high interest rate, and sale price well above the actual assessed value of the home. Prospective buyers pay a monthly rate, as with a typical mortgage, and often have to take on high home-repair costs and property taxes; however, they do not receive the legal title to the home until the full purchase price has been paid. If they default on a single payment, buyers forfeit their right to the property, with no equity. In many cases, prospective buyers of these homes are not aware of the terms under which they enter into such contracts, and Harbour has come under scrutiny in many locations for discriminatory lending practices.
Our task is to analyze property, demographic, and sales data for Harbour properties; visualize our findings through maps and an interactive web application; and, as appropriate, support the claim of discrimination for legal proceedings. We will trace the chain of ownership of Harbour homes over time, search for correlations between property locations and demographics, and create tools for Atlanta Legal Aid to more easily access the information they need.
We are also working on another project related to housing justice: Atlanta’s anti-displacement tax fund program. As neighborhood revitalization occurs in Atlanta’s Westside, there are concerns that longtime residents will be displaced as the area gentrifies. One proposed measure to decrease this turnover is the anti-displacement tax fund, which will offset the increase in property taxes for eligible residents. We will analyze these eligibility requirements to determine how many people qualify for the program, and think about ways in which this program or other anti-displacement measures could be expanded. We hope that the data analysis and tools we build will increase understanding of these two important housing justice issues.
Cycle Atlanta: Seeing Like a Bike
Seeing Like a Bike is a project under direction of Christopher A. Le Dantec, an Associate Professor of Digital Media in the School of Literature, Media, and Communication at Georgia Tech. The main purpose of this project is to identify environmental factors that affect the level of bike rider stress. During this summer, four students from different universities are working on the development of the bike-based sensor platform which will allows us to see what the bike sees. By the end of this term, we will have a small number of bikes mounting sensor array able to collect data to start to building predictable models of cycling stress base on environmental factors.
Building Energy Analytics
All Georgia Tech buildings have sensors which have been assessing how much energy is being used every 15 minutes for the past few years. Although a lot of data has been collected, for the most part, it has remained unutilized. As a result, our goal is to use that data, along with other external data like local weather and building occupancy, in order to model energy usage at Georgia Tech. In doing so, we hope to both determine which factors contribute to energy usage and target buildings that are most inefficient and in need of an upgrade.
UN Climate Challenge: Predicting & Alleviating Road Flooding in Senegal
Climate change will exacerbate existing socioeconomic vulnerabilities and threaten the success of crucial development schemes. Developing and maintaining resilient road networks is essential for meeting several UN sustainable development goals. However, currently about 75% of the road network in Africa is unpaved, making it especially susceptible to damage from precipitation.
As a result, our goal is twofold. One, we want to assess the effects of flooding events on road network connectivity in Senegal by determining which roads are most susceptible to flooding based on weather and topographical data. Two, we want to make recommendations for maintenance and upgrades that will enhance the climate resilience of the road network. We will do this by determining the volume of traffic of each road to quantify its contribution to accessibility between different parts of the countries, and perform a budget-constrained optimization framework to find a set of roads to target for improvements.
New American Pathways
New American Pathways is a new organization created on October 1, 2014, by the merger of Refugee Resettlement and Immigration Services of Atlanta (RRISA) and Refugee Family Services (RFS). Their goal is to resettle refugees coming to the United States and provide a range of services like literacy, employment, youth education and school services, and immigration and citizenship assistance. New American Pathways provides approximately 3,500 refugees per year with the necessary tools to rebuild their lives and achieve long-term success. So far, the primary area where refugees are resettled is Clarkston, GA. This area is understandably becoming saturated, and New American Pathways has partnered with DSSG to design a data-driven way to effectively identify other potential places for resettlement in the metro Atlanta areas. The main challenge is to find affordable housing within a mile of public transportation and near schools, faith centers and international grocery stores. Our team is working on designing an interactive tool that searches, filters and demonstrates a list of possible locations in Fulton and Dekalb counties that meet these criteria.
The scope of the project is quite interesting in that there are a host of facilities that refugee families need when they arrive in the USA. From being in an affordable area to having access to public transportation, easy access to shopping, grocery and health facilities and within easy access of the New American Pathways office which is currently based in Atlanta. Furthermore, refugee families also need English as Second Language classes, public schools with reduced lunch programs and government services such as DDS offices and DFACs offices nearby. All these constraints create an interesting but challenging problem, where our team is focused on displaying all the relevant information in one convenient and interactive tool, and potentially building a recommender which shows the client a list of top apartments based on the data of the refugee family arriving in Georgia.
Westside Communities Alliance
Data Science for Social Good is teaming up this summer with the Westside Communities Alliance (WCA), a network sponsored and managed by Georgia Tech to collaborate with neighboring communities, including English Avenue and Vine City. The mission of the alliance is to use the resources of Georgia Tech to collaborate and foster a vibrant and safe community.
Recently, the WCA unveiled its first interactive Data Dashboard that offers community leaders, researchers, local government agencies and foundations with data about the Westside communities in Atlanta. The data include historic and current records related to education and public safety, along with reports and other resources that could support data driven decisions and policies. The website aims to be a one-stop- shop for all things related to the Westside, and includes demographic and historic data about the community. It also has sections for transportation, a business directory, civic engagement, water and environment, some of which are still under development.
Our project is to create an interactive public safety mapping tool. The public safety module will fit into the existing framework of the data dashboard, and be made available via the existing dashboard website. This project aims to visualize the trends and the contexts of crimes in the Westside to give the community access to data and tools that have traditionally only been available to law enforcement agencies. By locating areas prone to criminal activity, and correlated data such as vacant housing, the community will be able use the tool to locate vulnerable areas and coordinate their own crime prevention strategies. In addition, our module will also examine data from the past in an effort to understand how criminal activity affects the most vulnerable of our citizens, the children and elderly. Ultimately, our philosophy is that data can be used to help tell the story of a community. Changes in the community are reflected in the data, and our goal is to connect residents to the data in a meaningful way.
Atlanta Regional Commission
The Atlanta Regional Commission (ARC) coordinates initiatives between the 10 counties and the cities within the Atlanta metro area. An important part of their work focuses on planning public transit between the many transit agencies throughout the region. Given the sprawling nature of Atlanta, public transit is critical to giving Atlantans access to the goods and services that they need. This summer, DSSG is collaborating with ARC to help them better analyze and assess the effectiveness of public transit throughout the city.
ARC is particularly interested in examining the intersection of public transit accessibility, employment, and poverty. Public transit serves as an essential tool for economic mobility to underserved communities, and ARC wants to be able to easily visualize transit accessibility to jobs from impoverished neighborhoods. Towards this end, we are developing a tool that allows users to view where they can get by walking or public transit for any given point in Atlanta. The tool will allow users to explore the number of jobs and types of jobs available in each range. It will also include a control for poverty so that ARC and other users can hone in on the neighborhoods that most need reliable access to jobs through public transit. We draw heavy inspiration from a similar tool created for New York City by their Regional Planning Association. We hope that our tool will enable ARC and members of the public to easily explore the intersection of public transit, employment, and poverty from any location in the Atlanta.
211 – United Way of Greater Atlanta
United Way of Metro Atlanta offers a serious commitment to their customers to maintain full satisfaction and ensure that all individuals have the opportunity to thrive and be part of a prospering community. The United Way of Metropolitan Atlanta was the first to introduce 211 services in 1997. It is completely operated by private a non-profit community-service organization. 2-1-1 handles a remarkable number of calls and provides informational services to large numbers of callers each day.
However; the 2-1-1 organization is overwhelmed with large numbers of help requests that cannot be handled in a speedy manner due to their limited resources. Can we find a way that enables the 2-1-1 staff to be able to handle every customer’s question or issue? And could some of the calls be answered via an automatic voice system?
This project aims to study 2-1-1 system and suggests different ways to improve and re-construct the existing call menu to better organize incoming calls to help reduce average wait time. Therefore; we are examining and analyzing all components to find ways of improvements. Some of those key components include; categorize typical time-consuming calls, identify the bottlenecks and problem areas that contribute to long wait times, and isolate sources of abandoned calls and explore ways to blacklist the numbers.
In addition to re-constructing the call tree, we are developing regression models to predict call volume, type of call, average talk time and peak call time in different periods of the day, month, and various seasons of the year. This will redefine the way 211 handles their calls and will provide the 211 management with the tools needed to consolidate staffing and resource assignment.
Fire Risk in Atlanta – Atlanta Fire Rescue Department
Hundreds of buildings catch on fire in Atlanta every year. The Atlanta Fire Rescue Department (AFRD) attempts to reduce fire risk by inspecting buildings for potential hazards and fire code violations, but they currently only inspect a subset of the total buildings needing inspection. How can AFRD become aware of more buildings in Atlanta with potential fire hazards, and how can they prioritize inspection to focus on those buildings with the greatest risk of fire? This is where DSSG comes in!
Our project aims to reduce fire risk in Atlanta by identifying and prioritizing buildings that should be inspected by AFRD. First, we use fire permit criteria and data about currently inspected buildings to identify more properties that should be inspected. Next, we use data about fires from the past five years to develop a predictive model for fire risk based on a property’s characteristics. Then, we prioritize the list of properties that should be inspected based on our fire risk prediction model, so that AFRD can use their inspection resources most effectively to reduce fire risk in Atlanta. Our final deliverable will be a web-based user interface that AFRD can use to interact with 1) the lists of buildings to inspect, and 2) information visualizations of fire-related Atlanta data. This partnership between AFRD and DSSG, which may substantially reduce fire risk in Atlanta, has the potential to save lives if successful and is an example of data science’s applications for real world impact! Project website: http://firebird.gatech.edu/
Trees Atlanta – Maintaining the Urban Forest
We are working with The City of Atlanta and Trees Atlanta to help them maintain and improve Atlanta’s urban forest. An urban forest is a forest or a collection of trees that grow within a city, town, or a suburb. Though Atlanta has one of the best urban forests in the country, it’s at risk of disappearing. Urban forests are an essential part of cities throughout the world; they provide many benefits to the city itself and its citizens. Some of these benefits include offsetting air pollution by trapping pollutants and absorbing CO2, mitigating the heat island effect, reducing surface runoff, increasing the property value of homes, and potentially reducing energy consumption by providing shade and serving as windbreaks in the winter.
A few challenges our partners are facing include efficiently identifying and choosing tree planting locations, identifying areas for forest preservation, and maintaining a diverse tree canopy. We are working on various data-intensive solutions for these problems.
Since Atlanta is such a large city, finding ideal planting locations can be challenging and time consuming. Using multiple types of data, such as percent tree canopy cover, impervious surfaces, and floodplain data, we are developing a model that prioritizes planting sites by land parcels. This will help quantify the benefits of planting trees in a given location, assist arborists in finding potential planting sites, and enable policy makers to make well-informed decisions about the future of Atlanta’s urban forest. The model will be presented in the form of a user-friendly web application that can function with little technical support.
One component of decision-making involves preserving large areas of forested land, some of which might be wildlife corridors. Identifying contiguous tracts of forest and related costs will help the Atlanta Tree Commission make decisions on which areas to conserve. This will also be part of the web application.
To withstand different diseases and pests, it’s important for an urban forest to have a diverse population of trees. Currently, arborists do not have access to visualizations of the tree population broken down by species within the city of Atlanta. We’re developing two interactive visualizations that will help arborists identify the biodiversity of Atlanta’s urban forest in detail.
We hope our work will help Atlanta support and expand its amazing urban forest. Project website: http://dssgtrees.gatech.edu/
Team WiFi is working with Georgia Tech and their campus WiFi data to understand patterns of mobility that can inform policymaking. This summer, we’re building a platform/tool to help condense, contextualize, and visualize this data for the benefit of others on campus.
WiFi data is more than just upload speeds or a network diagnostic tool; it’s a reliable indicator of user mobility and usage trends over space and time. Georgia Tech provides a secure WiFi network for its students, faculty, and staff. As a user moves around campus, their devices leave “breadcrumbs” at the WiFi access points nearby. These breadcrumbs give us a unique vantage point of campus – we have a bird’s eye and genuine view of broad behaviors and movement patterns throughout the day, week, and semester. We can see anomalies in day-to-day usage, like when the Snowpocalypse or the President visits campus. And, we can also learn more about buildings, occupancy rates, device connectivity, and other abstract information that cannot be inferred from other methods. This WiFi data provides the unique opportunity to learn about campus and give policymakers around Georgia Tech better data to inform their decision-making.
Our tool is the first attempt to give non-technical users a glimpse at these broad patterns of mobility. The first steps we’ve taken are to clean our data, condense the records, and filter out noise. Second, we abstract these breadcrumbs to broad features of movement around campus. Some of these features include movement within and between buildings, how users move to different spaces around campus, and how much time do people spend in specific locations. We then build a predictive model that clusters days into patterns of usage that we can then use to detect anomalous events. Our first attempts have successfully clustered days by class patterns (Monday-Wednesday versus Tuesday-Thursday versus weekends), and our model will provide other methods of clustering to help build a more accurate and robust system. Finally, we will create visualizations so that our model can display these trends.
Even though this WiFi data is a powerful start for learning about Georgia Tech, there are many well-grounded privacy, anonymity, and ethics concerns that arise from this kind of research. How fine-grain is the analysis, let alone what can be visualized? Can individuals be tracked? How do we protect the privacy of specific users or kinds of users while also using this kind of data for social good? A balance is needed between providing social benefits to communities and protecting the privacy of the users that provide this data to Georgia Tech. In addition to developing this tool, we are exploring different methods of data abstraction and generalizations that prevent reidentification and individual tracking while still creating the insights we mentioned above. We hope these insights will inform other researchers and campuses interested in using WiFi data to learn more about their campuses, companies, and areas.
City of Atlanta Emergency 911 Dispatch: Helping the city of Atlanta improve emergency response
Cycle Atlanta: Visualizing and Analyzing Cyclist Data for Safer Biking in Atlanta
Truly Living Well: Building Farm-Ready Tools to Collect, Manage, and Summarize Operational Data
Campus WiFi: Analyze WiFi authentication data from the Georgia Tech campus
Atlanta Community Court: One step forward in evaluation of an alternative criminal justice
City of Atlanta Emergency 911 Dispatch
Atlanta PD handles a truly impressive number of 911 calls each year (~1 million). These calls are handled by a group of call takers, and subsequently dispatchers that are each responsible for a specific zone in the City. During a normal shift, a dispatcher will cover the calls requiring dispatches in their zone and prioritize a police unit from the nearest beat as quickly as possible. It is a very demanding job, and the dispatchers have to take into account the priority of the call and what calls the police units are currently assigned to. The city of Atlanta employs around two thousand police officers, and the dispatchers help keep them distributed throughout the city at any given moment to not only respond to crimes, but to also provide a visible police presence that deters crime from occurring in the first place.
The project aims to help Atlanta better understand how the workload of dispatching and responding to calls is currently distributed throughout the city. Through this process we are redefining what dispatcher workload means and developing several alternative techniques and scenarios for altering this distribution to improve dispatch times. We are developing web-based visualization tools that help the dispatchers explore the data by identifying the locations where police are most frequently dispatched, the response times of those locations, and the types of calls that most commonly occur. These tools will help the police leadership develop new strategies for improving dispatch and response time when responding to 911 calls.
Truly Living Well – Building FarmReady Tools to Collect, Manage, and Summarize Operational Data
In large cities across the country, “food deserts” inevitably emerge – poor areas with limited access to fresh produce. In response, many communities are turning empty lots into smallscale urban farms as a source of healthier, sustainable food. Unfortunately, actually running a miniaturized farm that meets the needs of local citizens is a complex undertaking with many inputs and outputs to manage, and inefficiency creeps in. Effectively managing available farm data would go a long way towards increasing productivity, but the onus of recording information typically ends up unduly burdening farmers and the data collection dries up. This means that even as nonprofits like Truly Living Well experience explosive growth and support from their communities, apparently straightforward operational information such as visitor demographics, timecards, or harvest requests can be difficult or impossible for them to gauge accurately.
By working with local farmers to best assess their immediate needs, our group is putting together the first available web application specifically designed for data collection, management, and analysis at the scale of urban farming. By helping farmers to visualize the extent of volunteer support, their upcoming harvest requests, and sales patterns in their communities, we can help urban farming to function more efficiently and continue growing in a sustainable, informed way.
Atlanta is known as a carcentered city, but it is also home to a steadily growing cycling community. These two characteristics don’t always mix, and this raises concerns about cyclist safety throughout the city. The Cycle Atlanta mobile app was built to help address these very concerns. Developed by the Participatory Publics Lab at Georgia Tech, Cycle Atlanta allows cyclists to track their rides throughout the city. In addition to tracking their own riding patterns, app users also provide valuable information to city officials for planning future cycling infrastructure throughout Atlanta.
However, collecting this data is only part of the process. In order to gain better insight into where, when, and how Atlanta’s cyclists ride, the data needs to be cleaned, anonymized, and analyzed. To this end, our project has several goals including development of interactive visualization tools for the data currently being collected. These tools will allow bike enthusiasts and the public to explore data collected by the app while also providing stakeholders an opportunity for future exploratory analysis. Additionally we’re working to engage with the cycling community within Atlanta to gain feedback on the current implementation of the mobile app. This will aid our group in providing recommendations for future work on the app.
Atlanta Community Court
If a room in your home is messy, would you tear it down, or would you clean it? Most people would agree cleaning is a better choice, since it protects the structure of the house. The Community Court of Atlanta protects the structure of society in the same way by providing restorative justice, which is a less destructive alternative to incarceration. The indigent and mentally ill receive proper treatment and care in the Community Court, instead of going straight to jail. In doing so, the Court cuts the cycle of incarceration and helps bring offenders back to the mainstream of society.
We’re working with the Community Court to produce an evaluation of the effectiveness of their programs. As a first step towards this goal, the project team is simultaneously cleaning the existing raw data and constructing a standardized framework of data collection. Geographic and demographic statistics of the transformed data will be reported, as well as templates of surveys and a data input system. Using these templates, temporal and snapshot data collected in the future will enable development of advanced models to study the effectiveness of the programs.
Georgia Tech IT
We are working to analyze a large data set consisting of authentication logs from the Georgia Tech campus. This consists of data authentication times and locations for users over a one year period from the CLOUGH building. We are analyzing the clustering and migration patterns for different demographics and time periods. We are also investigating various models to predict occupancy.