2018 Capstone Projects
Prediction of Mode Shift in Cities Based on Trip Cost-Duration Manhattan District
CUSP Mentors: Kaan Ozbay & Abdullah Kurkcu
CUSP Students: Ian Xuan Xiao, Benjamin Alpert, Tyler Woebkenberg, Unisse Chua, Andrew Nell
With the introduction of shared mobility services and the promise of autonomous vehicles, little is known about any anticipated changes in consumer mode preferences for urban mobility services resulting from shifts in affordability and accessibility. This is particularly true for understanding the impact to ridership of public transit systems across the United States, which are seeing declines in ridership while app-based ride hailing services are experiencing rapid growth.
This study focuses on understanding the commuter hour (6:00 AM to 10:00 AM) ridership trends of subway and app-based ride-hailing services (specifically Uber) and features that influence them. New York City was chosen as the initial city for the project due to the richness of data. While there are still shifts occurring in market share for both Uber and subway, most parameters have remained constant suggesting the system is not in equilibrium after the disruption of Uber. In order to understand and predict outcomes in the event of a price shift, both the longitudinal forecasting (without changes in features) and a cross-sectional study (features influences at a fixed point in time) need to be studied and incorporated into a model.
Public Sentiment Measures Towards Police from Social Media
Sponsor: BetaGov @ NYU
CUSP Mentor: Federica Bianco
CUSP Students: Yuwen Chang, Yukun Wan, Scott Smith, Gaurav Bhardwaj
Community support plays a crucial role in the effectiveness of prosecutorial justice sector and law enforcement. However, measuring public sentiments towards relevant policy implementations is difficult. Conventional methods are labor-intensive, time-lagged, and not scalable. Alternatively, we investigate the viability and strategies of leveraging social network services to assess public sentiments in a massive, timely, and effective manner.
Focusing on Twitter, we first evaluate strategies to harvest relevant tweets and mitigate contamination. This includes keywords selection and streaming pipeline modification. Then, using the collected tweets regarding prosecutorial justice departments at both local and national levels, we assess whether a sentiment baseline exists based on the positive, neutral, or negative sentiments perceived by existing Natural Language Processing models. Finally, we investigate geographical and longitudinal variations in sentiments.
NYC311 One Stop Shop
Sponsor: CUSP Data Facility and NYU Furman Center
CUSP Mentors: Julia Lane, Clayton Hunter, Ingrid Gould Ellen, & Xiaodi Li
CUSP Students: Franz Loza, Sarah Schoengold, Baiyue Cao, Sofiya Elyukin
Project Website: https://www.nyc311onestopshop.com/
NYC311’s mission is to provide the public with quick, easy access to all New York City government services and information. The NYC311 dataset has been called the pulse of the city, and can add texture to research questions in public policy and civic technology fields. While the data set is available for free online on NYC’s Open Data Portal, it is both difficult to use for analysis and challenging to interpret.
Using a Human-Centered Design approach to identify the overlapping priority needs of the diverse NYC311 data users, we developed a one-stop shop site with three core features:
- A 311 download tool to quickly access the data in a useable format
- A user manual to better understand where the data comes from
- A compilation of external resources to explore 311-related topics
Digital Traces of Gentrification
Sponsor: CUSP Urban Complexity Group
CUSP Mentor: Stanislav Sobolevsky
CUSP Students: Juan Sokoloff, Baoling Zhou, Srikanth Namuduri, Lingyi Zhang
Project Website: https://srikanth261.wixsite.com/cusp
New York City has a lot of neighborhoods that are rapidly changing, generating a transformation of the urban environment and economic activities in them. But it is not all good; this renovation also generates an increase in the average rent prices and potentially the displacement of low-income people who live in those. This process is called gentrification.
In this project we wanted to find out if there is a way to know where gentrification is happening based on the indicators of human behavior. So we took a bunch of open-source data and developed some target variables for gentrification. And then related these to human behavior data. After running some boring but interesting statistical techniques we find that there are actually some variables that can help explain gentrification.
Measuring the Impact of Flooding Risk on Affected Housing Market
Sponsor: NYC Department of City Planning
CUSP Mentor: Neil Kleiman
CUSP Students: Hans He, Yu Chen, Chenrui Zhang, Emily Padvorac
Project Website: https://ch3183.wixsite.com/capstone2018
Six years after Superstorm Sandy, how is the housing market in flood zone areas doing? Does the flood risk impact them and their neighboring areas? What are the factors that impact the housing market in flood zone areas? Can we predict and identify the change in housing prices?
This study targets 1-3 family houses within New York City’s five boroughs that are located in the 100 year flood zone areas. This study uses Federal Emergency Management Agency’s (FEMA) legacy flood maps FIRM07, and the recently updates flood map PFIRM15.
NYC census tracks will be used to group flood zone housing markets. We will use machine learning techniques to find important factors that impact housing market prices, and to cluster flood zones that have similar housing market trends and analyze the impacts of flood risk on neighborhoods in non-flood zone housing markets. We also developed a Kriging model to predict flood zone housing prices and identify property changes.
Impact of Artificial Light at Night on Bird Migration
Sponsor: New York City Audubon
CUSP Mentor: Gregory Dobler
CUSP Students: Matthew Dwyer, Emily Hansen, Christian Moscardi, Cheng Ma
Project Website: https://cm4692.wixsite.com/light-birds
Millions of birds are killed annually in urban areas from crashing into buildings or being trapped by artificial light at night (ALAN). This is especially fatal during migration seasons in the fall and spring, with a demonstrated association between light and flight paths. The Urban Observatory at NYU’s Center for Urban Science and Progress utilizes large-scale sensing and computing capabilities to observe and analyze the dynamics of urban environments. The data for this study came in part from the UO’s pair of cameras stationed in One Bryant Park in Manhattan. The sponsor for this project, New York City Audubon, is a grassroots community that tirelessly works for the protection of wild birds and wildlife in and around New York City.
Predicting Illegal Gas Conversions
Sponsor: NYC Department of Buildings
CUSP Mentors: Justin Salamon & Charlie Mydlarz
CUSP Students: Akkas Ozgur, Kelsey Reid, Nicholas Jones, Hongkai He
Project Website: https://www.dobpredictions.com/
In the wake of a 2015 gas explosion in Manhattan, the New York City Department of Buildings (DOB) has been working to better allocate its inspectors in an effort to reduce the risk of future deaths and property damage resulting from illegal gas plumbing.
With a finite number of inspectors and a continuous stream of gas related complaints, the DOB needs to develop a way to better review, prioritize, and investigate these complaints to issue violations and stop illegal gas work before another property and life-destroying incident occurs.
Using DOB-provided inspection data, along with other data sets of building and neighborhood-level phenomena in the city, this work seeks to develop a risk model for deployment within DOB. For a new complaint received, the model will provide a risk score representing the likelihood of finding illegal or dangerous gas-related conditions at that address.
The project team developed and validated a machine learning algorithm that assigns unique outcomes to complaint histories; engineered a dataset with attributes using the past DOB inspections, PLUTO land use and assessment, Housing and Preservation, FDNY and Census data; identified strong prediction indicators and built a Random Forest machine learning model to produce violation risk scores at building level. This model will provide an effective inspection decision-support tool allowing the DOB to apprehend more cases of illegal gas alterations and support the Department’s initiatives to improve gas plumbing safety in New York City.
Property Valuation & Tax Mapping from Imagery Data
Sponsor: NYC Department of Finance
CUSP Mentor: Daniel Neill
CUSP Students: Zhiao Zhou, Nina Nurrahmawati, Bailey Griswold, Te Du
Automated tax valuation models utilize individual building features to estimate a home’s value and subsequent tax liability. The records that the New York City Department of Finance (DoF) has that document the features used for tax assessment for each house currently have no quality assurance checks apart from in-person inspections done by visits to each individual home. Desktop review of high resolution street level images is an effective replacement for on-site inspections, but still requires manual labor, which, multiplied over the more than 1 million parcels in the DoF’s jurisdiction, poses a large drain on resources
We want to test the feasibility of an automated approach of home-feature screening. Automated home-feature detection is a scalable technique, but it is only worthwhile if it is effective. Therefore, we must prove the feasibility of this approach by training an accurate home-feature image classifier. The other goal of this study is to compare the performance of classifiers with training images of varying quantities and qualities. While image classifiers perform best when trained on a large volume of high quality images, labelling those training images is another resource cost, and requires the desktop review that this approach is meant to replace. In theory, the DoF already has labels for every address, since they have a library of features for every house. Although some of those labels are wrong, we test the performance cost of using these “noisy” labels as training labels.
Vulnerability of Transportation Networks 2.0
Sponsor: CUSP Urban Complexity Group; co-sponsored by Lockheed Martin Advanced Technology Labs
CUSP Mentor: Stanislav Sobolevsky
CUSP Students: Alexander Shannon, Rachel Lim, Sunglyoung Kim, Fangshu Lin
Subway systems form the backbone of the city, ensuring the smooth functioning of daily activities. The scale of subway systems in these cities make them vulnerable to disruptions. Hence it is crucial for cities to anticipate repercussions to maximize the effectiveness of the system. While single node failure has been relatively well studied, disruptions on multiple nodes have been relatively unexplored. Our study is driven by the idea that simultaneous disruption on two stations may have more significant impact than the cumulative disruption on two separate stations, a situation likely to occur during a natural disaster or terrorist attack. We present the idea of synergistic effects, and by comparing network disruption in 5 cities, we uncover the patterns and universalities between subway systems worldwide
Digital Traces of Tenant Intimidation
Sponsor: CUSP Urban Modeling Group
CUSP Mentor: Debra Laefer
CUSP Students: John Lundquist, Yixuan Tang, Keith Dumanski, Julian Ferreiro
Landlords of residential properties have unprecedented influence in New York City. A growing demand for apartments and a gradual weakening of tenant protection laws in New York have resulted in an immense financial incentive to displace rent-stabilized tenants by whatever means necessary. Regulators do not have the capacity to keep up, meaning that landlords are often not held accountable for the harm they cause through the predatory measures taken displace rent stabilized tenants.
Our Mission: use data to identify, characterize and hold to account New York City’s predatory landlords.
Sidewalk Bridges: The Unregulated Eyesore
Sponsor: CUSP Urban Modeling Group
CUSP Mentor: Debra Laefer
CUSP Students: Marc Toneatto, Rebecca Scheidegger, Jonathan Geis, Shreya Bamne
In New York City, sidewalk sheds are deployed for new construction, building alterations, and as part of Local Law 11 Facade Inspection regulations. Local media, politicians, and citizens commonly comment on the negative impacts sheds have on the local community. However, it has not been quantified or empirically measured if sheds actually adversely impact the community. Through considering 311 complaints and NYPD crime data, it is explored if there is evidence to support the claim that sheds have negative impacts on communities. While it is not found that there is conclusive evidence that sidewalk sheds consistently have negative impact communities, improved data sources would likely allow for a more nuanced and concrete answer. To aid in facilitating NYC Department of Buildings work of overseeing sheds and prioritizing what sheds are most in need of attention, complaint and violation data is combined from multiple sources to create a prototype ranking system that highlights sheds that may be problematic or in need of intervention.
Detection of Polluting Plumes Ejected from NYC buildings
Sponsor: CUSP Urban Observatory
CUSP Mentors: Federica Bianco & Gregory Dobler
CUSP Students: Chun-Chieh Tsai, Anupama Santhosh, Benjamin Steers, Jonathan Kastelan
Detecting, counting, and tracking instances of plumes ejected from NYC buildings to help gain a better understanding of the building metabolism and the environmental and health impacts of building energy use.
A plume is a cloud of smoke or vapor emitted from buildings during the production of energy, through the burning of heating oils. They are typically made up of pollutants such as PM 2.5, SO2, and CO2. NYC has ~10,000 buildings that are still burning heating oils.
75% of greenhouse gas emissions are from buildings, according to the City of New York. If we want to be able to reduce emissions, we need to understand the emission patterns of buildings. We trained Faster R-CNN, an image classification and localization model, to detect plumes automatically using RGB images taken from a building-top camera. It provides a cost-efficient way of measuring the frequency of building emissions without the need of deploying an expansive in-situ sensor network to detect plumes using local air quality measurements.
Map of Gentrification and Displacement for Greater New York
Sponsor: Local Initiatives Support Corporation
CUSP Mentor: Neil Kleiman
CUSP Students: Ruben Hambardzumyan, Dana Chermesh Reshef, Gerardo Rodriguez Vazquez, Hao Xi
Project Website: http://www.udpny.org/
While gentrification can bring new resources to neighborhoods, however, it also can result in more negative transformation of urban neighborhood landscapes, in particular contributing to the displacement of low-income residents. By utilizing the methodology created by the Urban Displacement Project to analyze the neighborhoods of San Francisco Bay Area, this capstone project explores gentrification and displacement in the areas of the Greater New York, mainly through the lenses of population displacement and exclusion. Via an extensive analysis of socio-demographic characteristics obtained mainly from the American Community Survey (ACS five-year estimates), and collaborations with different stakeholders from both the academia and practice, we show that processes of gentrification, displacement, and exclusion are taking place throughout the 31-county region.
This NYU CUSP capstone project seeks to support research on the Greater New York housing market and residents, as well as to contribute to the US and worldwide discourse on gentrification and displacement.
Automated Detection of Street-Level Product Displays
Sponsor: NYU mHealth
CUSP Mentors: Gregory Dobler, Federica Bianco & Thomas Kirchner
CUSP Students: Prince Abunku, Isha Chaturvedi, Jianghao Zhu, Guobing Chen, Charles Moffett
Tobacco marketing, restricted almost exclusively to the point-of-sale in recent years, is extremely effective in getting more people to consume and fewer to quit these deadly products. The lack of empirical documentation linking product exposure to behavior, however, is a key obstacle to the adoption of additional restrictions on point-of-sale tobacco advertising. The goal of this project is to map point-of-sale tobacco marketing practices across New York City using automated detection of tobacco signage in street-level imaging data. We propose to build and train convolutional neural networks, which are particularly effective at detecting objects in images, to identify and classify outdoor advertisements of cigarettes and smokeless tobacco. Previous analyses of visual data in public health research involving manual image coding, though made more efficient over time with the help of crowdsourcing, are prohibitively costly and time-consuming. The importance and motivation of the project stems from the immediate and comprehensive effect of tobacco advertisements on its sales and consequently on public health. Once the model is developed, it can be used in measuring exposure of at-risk communities to tobacco displays.