You are on page 1of 76

AdventuresinUrbanInformatics

UniversityofCalifornia
February,2016
Dr.StevenE.Koonin,CUSPDirector
steven.koonin@nyu.edu
http://cusp.nyu.edu

BigCities+BigData

Theworldisurbanizing
Citiesarethelociof
consumption,economic
activity,andinnovation

Citiesarethecauseofourproblems
and thesourceofthesolutions

Allcitiesmustbebetterforglobalissues
Individualcitiesneedtobebestforcompetitivenessin
talent,capital,
Beefficient,resilient,sustainable
Addresscitizenqualityoflife,equity,engagement

BigCities+BigData
Informatics
capabilitiesare
exploding
Storage,transmission,
analysis

Proliferationofstatic
andmobilesensors
Internetofthings
Globalnetworktraffic,30%CAGR

JUSTHOWDIDAPHYSICISTWIND
UPINTHISBUSINESS?

Whatdoesitmeantoinstrumentacity?
Infrastructure

Environment

People

Condition,operations

Meteorology,pollution,
noise,flora,fauna

Relationships,location,
economic/communications
activities,health,nutrition,
opinions,organizations,

Properlyacquired,integrated,andanalyzed,datacan

Takegovernmentbeyondimperfectunderstanding
Better(andmoreefficient)operations,betterplanning,betterpolicy

Improvegovernanceandcitizenengagement
Enabletheprivatesectortodevelopnewservicesforcitizens,
governments,firms
Enablearevolutioninthesocialsciences

UrbanData
Urbandatahavebeencollectedformillennia
statistics (sttstks)n. 1.Themathematicsofthecollection,organization,and
interpretationofnumericaldata,especiallytheanalysisofpopulation
characteristicsbyinferencefromsampling
FromGermanStatistik,politicalscience,fromNewLatinstatisticus,ofstateaffairs,from
Italianstatista,personskilledinstatecraft,fromstato,state,fromOldItalian,fromLatin
status,position,formofgovernment.

Sparsenessandqualityhavelimitedurbanscience
difficulttousefullymeasuretheurbansystem,testhypotheses

Butnewdatatechnologiescompletelyrecastthestudyofcities
digitalrecords,sensors,computingpower,analyticaltechniques
unprecedentedgranularity,variety,coverage,andtimeliness
Whenyoucanmeasurewhatyouarespeakingabout,andexpressitinnumbers,
youknowsomethingaboutit;whenyoucannotexpressitinnumbers,yourknowledge
isofameagerandunsatisfactorykind. LordKelvin,1883

UrbanDataSources
Organicdataflows

Administrativerecords(census,permits,)
Transactions(sales,communications,)
Operational(traffic,transit,utilities,healthsystem,)
Twitterfeeds,blogposts,Facebook,

Sensors

Personal(location,activity,physiological)
Fixed insitusensors
Crowdsourcing(mobilephones,)
Chokepoints(people,vehicles)

Opportunitiesfornovelsensortechnologies

Visible,infraredandspectralimagery
RADAR,LIDAR
Gravityandmagnetic
Seismic,acoustic
Ionizingradiation,biological,chemical

BuildingEnergyEfficiency
Source Energy Use Intensity, Office Buildings, New York City

Source Energy Use Intensity, Multi-Family Buildings, New York City


Source: Local Law 84 Energy Disclosure Data, Kontokosta 2013

800

80

Source: Local Law 84 Disclosure Data, Kontokosta 2013

N = 1,150
Mean = 219.5

N = 7,505
Mean = 137.9
s.d. = 46.8

20

200

Fre qu en cy
40

Frequency
400
600

60

s.d. = 101.7

200
400
600
Weather Normalized Source EUI (kBtu/sq.ft./yr.)

Kontokosta 2013

800

100
200
300
Weather Normalized Source EUI (kBtu/sq.ft./yr.)

400

LocalLaw84BenchmarkingData

Kontokosta,2013

CellTowerRecordsforTrafficAnalysis

Wang,P.,Hunter,T.,Bayen,A.M.,Schechtner,K.&Gonzalez,M.C.
UnderstandingRoadUsagePatternsinUrbanAreas.Nature,Sci.Rep.2,1001;DOI:10.1038/srep01001(2012).

Dailycommutepatterns
fromphonerecords

Survey Chicago,Paris
Phone 4X104 inParis
Model Chicago,Paris

TaxisasSensorsforNYC
Taxisaresensors thatcanprovide
unprecedentedinsightintocitylife:economic
activity,humanbehavior,mobilitypatterns,
WhatistheaveragetriptimefromMidtowntotheairportsduringweekdays?'
Howthetaxifleetactivityvariesduringweekdays?
HowwasthetaxiactivityinMidtownaffectedduringapresidentialvisit?'
HowdidthemovementpatternschangeduringSandy?
Wherearethepopularnightspots?

StudyingTaxiPatterns

TrainStations
Airports

May1st 7th
2011
3.6MillionTrips

TaxiRidesinManhattan,October28 November3,2012
(Superstorm Sandy)

JulianaFreire,ClaudioSilva,etal,NYUPoly

Lottery Vis

Correlatesaleswith
Weather
Sportsteamwins
Twittermood

FromtheWillisTower,Chicago

ManhattanintheThermalIR
199WaterStreet
Built1993::998,000sqft
electricity,naturalgas,steam
LEEDCertified

PhotobyTyroneTurner/NationalGeographic

Othersynopticmodalities:Hyperspectral,RADAR,LIDAR,

The view from CUSPs Urban Observatory in Brooklyn

Borough Block & Lots (BBL)


Standard UO view
colored by distance

Picturemergesimagecapturedfromvideo,3DLIDARmapofNYC,PLUTO
(PrimaryLandUseTaxLotOutput)database,andLL84EnergyBenchmarkingdata
Source:Dobler,etal.

21

City Lights Project


OBJECTIVES

Develop a fundamentally new modality for


studying the city from a distance

Identify aggregate patterns of light in the


time-dependent brightnesses of city lights

Leverage these patterns into foundational


contributions to urban science and urban
functioning

Proof of Concept
IMPACT
Urban Science

Determine the underlying drivers of the pulse of the city


Dobleretal.;

Understand the effects of perturbations


doi:10.1016/j.is.2015.06.002
City Life

Monitor energy consumption by proxy using light patterns as a measure of building


occupancy

Evaluate the effects of disturbances (e.g., light/noise pollution) on public health

Camera:

Point Grey Flea 3 USB ; 8.8 Mega-pixels ; Raw image output ; 25mm focal length lens

Observations:

1 image every 10 seconds from Oct 26 to Nov 16, 2013; 3 color images at 25MB each
Total data volume ~4.5TB; Custom data processing pipeline

Privacy Protections
Institutional Review Board
approval of all projects involving
non-open data
Close oversight by CUSP Chief
Data Officer
Limited # of pixels per window
(but atmosphere/instrument
effects typically dominate)
Aggregate and de-identified
analysis only

Dynamics of the Urban Landscape

Each frame is registered to a common frame by spatial correlations


4,200 window apertures are identified by hand
(out of approximately 20,000 windows in the scene)

For each frame, the average brightness of each source is calculated in


3 bands (RGB)
The brightness of a given source as a function of time is referred to
as its light curve

Pulse of the City Lights

Daytime Phenomenology

11:00 AM

11:01 AM

Daytime Phenomenology: Subtle Variations

Daytime Phenomenology: Subtle Variations (animation)

raw image

Subtracting the City


Background subtraction:

background subtracted

registration to reference image


form 10 absolute difference images from
surrounding frames
construct the minimum difference image pixel by
pixel

Crossbuildingviewofaboilerplume
Suchplumesmaynotbevisiblefromstreetlevel.

raw image

Plumes of Opportunity
Background subtraction:

registration to reference image


form 10 absolute difference images from
surrounding frames
construct the minimum difference image pixel by
pixel

Plume identification and tracking:

denoise background subtracted image


identify excess/deficit in luminosity space
cross check object location in color space
localization and probability weighted tracking of
centroids

Upcoming use cases:

background subtracted

plume rate
repeaters
urban winds
carbon vs steam emissions
TOO (triggered) observations

Thoughtsonthebigsciencequestions
Canwedocumentthepulseofthecityinitsvariousdimensions?

Normal?Variability?Correlations?
Responsetoperturbations?Predictability?Precursors?

Howdothemacroobservablesarisefrommicrobehavior?

SantaFescaling?
Physicalstructureofcities?
Decisionrulesinagentbasedmodels
Roleofgeography?Culture?Policies?

TheCUSPPartnership
UniversityPartners

NationalLaboratories

NYU/NYUPoly
UniversityofToronto
UniversityofWarwick
CUNY
IITBombay
CarnegieMellonUniversity

LawrenceLivermore
LosAlamos
Sandia
Brookhaven

IndustrialPartners

City&StateAgencyPartners

IBM
Microsoft
Xerox

TheCityofNewYork

Cisco
ConEdison
Lutron

NationalGrid
Siemens

AECOM,Arup,IDEO

Buildings
CityPlanning
CitywideAdministrative
Services
DesignandConstruction
EconomicDevelopment
EnvironmentalProtection
Finance

FireDepartment
HealthandMentalHygiene
InformationTechnology
andTelecommunications
ParksandRecreation
PoliceDepartment
Sanitation
Transportation

MetropolitanTransitAuthority
PortAuthorityofNY&NJ

Education Programs at CUSP


Master of Science in Applied Urban
Science & Informatics
F/T (One Year)
P/T (Two Year)

Civics Analytics Track


Advanced Certificate in Applied Urban
Science & Informatics

Graduate Programs at CUSP

Interdisciplinary and cutting edge approach that links data science,


statistics and analytics, and mathematics with complex urban
systems, urban management, and policy.

Core
curriculum

Tracks

Length

Urban core

Foundational understanding of the theories of urban planning and


the application of data-driven approaches to urban challenges.

Informatics core

Fundamentals of data science/computer science, data management,


data mining, visualization, model selection, and machine learning
tools to urban problems and datasets.

Urban
Informatics

For students who are looking for deep training in data science and
informatics as applied to cities.

Civic Analytics

For students who will utilize analytics and data-driven decisionmaking techniques to inform urban operations and policy decisions.

One Year FullTime Program

A research- and project-intensive environment

Two Year PartTime for


Working
Professionals

Evening courses with numerous opportunities for networking with


peers, faculty, and experts in the industry.

CENTER FOR URBAN


SCIENCE+PROGRESS

MSAUSI - URBAN INFORMATICS TRACK (One Year)

Year 1

PRE-FALL

FALL

SPRING

SUMMER

1001 Urban
Computing Skills Lab

5003 Principles of
Urban Informatics

5006 Machine
Learning for Cities

6004 Advanced Topics


in Urban Informatics

1000 City Challenge


Week

5004 Applied Data


Science

9001 Urban Science


Intensive I: City
Operations & Applied
Informatics

9002 Urban Science


Intensive II: Practicum

4001 Computational
Urban Policy &
Planning

Data Science Elective

1007 Data
Governance, Ethics,
and Privacy

Select 1 from:

Domain Application
Elective

7007 Urban Spatial


Analytics

Informatics Core

9002 Urban Decision


Models

6001 Science of Cities


Research Seminar
6003 Civic Technology
Strategy

Urban Core
Optional Courses
Winter
Week

CENTER FOR URBAN


SCIENCE+PROGRESS

Select 1 from:

Spring Break
Data Dive

Admissions Summary, Class of 2014


Inaugural Academic Year: September 2013 July 2014

24

21%

27

36%

3.5

Inaugural Class

Selectivity

Years
Average Age

Female

Average
Undergraduate GPA

20

48%

28%

Undergraduate
Disciplines

International

Countries
Represented

Years Average
Work Experience

With Graduate Degree

(including 1 Adv. Cert.)

CENTER FOR URBAN


SCIENCE+PROGRESS

2015
ClassCohort
Highlights
Fall 2014

19

32%

COUNTRIES
(111% )

GRAD
DEGREES
completed/in-progress
CENTER FOR URBAN
SCIENCE+PROGRESS

45%

FEMALE
(275% )

YRS. AVG. WORK


EXPERIENCE
(25% )

28

AVERAGE
AGE (3% )

NYC
EMPLOYEES

Fall 2015 Cohort

87

NEW STUDENTS
August 2015

Fall 2015 Cohort

10

NYC Employees
August 2015

StudentResearch(GRAProgram,19projects)
Aerial Analytics/Hyperspectral imaging
Bluetooth Tracking Researcher
Buildings Informatics Energy Index
Dynamics of the City Lights
Economic Impacts of Public Parks and Greenspaces
Efficiently Indexing the New York City Open Data For Spatialtemporal-keyword Queries
Emotion Sensing
Garbage Identification
Machine Learning and Computational Statistics for NYPD
MTA Project
Parks Utilization and Attendance
Pedestrians and Vehicles: Interactions at Intersections
Quantified Community Research Initiative
Social Cities Initiative
SONYC
TaxiVis
Traffic Safety from Video Recordings
Understanding the spatial structure of crime
Urban Thermodynamics
Using MTA bus-time data to determine traffic conditions

2015CapstoneProjects

ACCESSNYC DataAnalysis
AnalysisofCitibike DataandModelingofTimeDependentOrigin
DestinationMatrices
Building&SustainabilityInformatics
BusVis:InteractiveExplorationofNYCBusData
CrimeandPolicingAnalyticsinNewYorkCity
DigitalEquality:Sensing,citizenscience,dataanalytics&
visualization
FromLightVariabilitytoEnergyConsumption
LearnrASeamlessEducationVolunteeringPlatform
NewYorkCityEconomicMap
NewYorkOpenGovernment
ParksQualityAssessment
QuantifyingParticulateMatterExposure DistributioninNYC
QuantitativeAnalysesofUrbanTopography
UrbanWasteAnalytics
UsingSocialMediatoPredictUrbanTransportation

Alumni Profiles
Career Changer
Alex Chohlas-Wood (B.A. Studio Arts)
Director of Research & Evaluation, NYPD

Become a Data Scientist


Warren Reed (B.S. Chemical Engineering)
Data Scientist, Office Of Financial Research

Advance your Career


Aliya Merali (B.S. Physics)
Director of Learning and Access, Coalition For Queens

CENTER FOR URBAN


SCIENCE+PROGRESS

CUSPFacilities/Capabilities
UnderDevelopment
Datafacility
QuantifiedCommunity
SONYCproject
UrbanObservatory

DataFacility
Overview
OmnivorousingestiontoarepositoryforNYCrelateddata

ObjectiveandGoals
Makedatainteroperable,withpropermultilayeredaccessprotocols

Data
DatafromCityagenciesonoperations,schedules,maps,etc.
WorkingwiththeMayorsOfficeofAnalytics
Startwiththeopendatasets
Includesproprietarydata,socialmediadata,CUSPgenerateddata

CUSPChiefDataOfficeroverseesethical,legal,andsocialissues
https://datahub.cusp.nyu.edu/dataset

NYCDataBridge

Privacy, Big Data, and the Public Good:


Frameworks for Engagement
The book identifies ways in which vast new sets of data on human beings can
be collected, integrated, and analyzed to improve urban systems and quality of
life while protecting confidentiality. Sponsored by CUSP, the American Statistical
Association, its Privacy and Confidentiality subcommittee, and the Research
Data Centre of the German Federal Employment Agency.
Editors: Julia Lane, American Institutes for Research; Victoria Stodden, Columbia;
Stefan Bender, The German Federal Employment Agency; Helen Nissenbaum, NYU

Chapter Authors
Steve Koonin, CUSP; Frauke Kreuter, U-MD and Richard Peng, Johns Hopkins; Alessandro Acquisti, Carnegie
Mellon University; Robert Goerge, UChicago; Helen Nissenbaum, NYU; Kathy Strandberg, NYU;
Paul Ohm, Colorado; Victoria Stodden, Columbia; Alan Karr, National Institute of Statistical Sciences and
Jerry Reiter, Duke University; John Wilbanks, Sage Bionetworks/Kauffman Foundation;
Cynthia Dwork, Microsoft; Alexander Pentland, et al., MIT; Carl Landwehr, George Washington
University; Peter Elias, University of Warwick.

The Quantified Community


Understanding the Patterns of Urban Life
The CUSP Quantified Community (QC) will be
a fully instrumented urban neighborhood that uses
an integrated, expandable sensor network and
citizen engagement to support the measurement,
integration, and analysis of neighborhood
conditions.
Through an informatics overlay, data on physical
and environmental conditions and use patterns will
be processed in real-time to maximize
operational efficiencies, improve quality of life
for residents and visitors, and drive evidencebased planning.

Kontokosta, et al.

Buildings
Resource consumption;
indoor air quality;
productivity, health
measures

Infrastructure
Solid waste, storm-water
management, power
generation/distribution

Safety and Security


Network Security,
Situational Awareness,
Emergency Management
Integration, Event
Forecasting

Environment
carbon emissions; air
pollution and particulates;
noise; climate

People
Behavior; mobility;
health; activity; social
networks, metagenomics

SoundofNewYorkCity
Cyberphysicalsystem forlargescale,continuous
monitoringofnoisepollution
Customacousticsensor (~$100/unit),dBmeasurement
accuratetocityagencystandards(+/2dB)
Stateoftheartmachinelistening technologyfor
automaticsoundsourceidentificationinrealtime
Alsoincludescitizenscience&datavisualization
components
J.Bello,C.Mydlarz,J.Salamon
54

Cyberphysicalsystemforlargescale
continuousmonitoringofnoisepollution

55

CustomacousticsensorbasedonMEMS
microphonetechnology

56

Stateoftheartmachinelisteningtechnology
forrealtimesoundsourceidentification

57

UrbanObservatory
Provisionedurbanvantagepoint(s)include

DowntownBrooklyn
MidtownManhattan

Suiteofboresightedinstruments

Photometricandcolorimetricoptical imaging
BroadbandIRimaging(SWIR,MWIR,andthermal)
Hyperspectralimaging(tracegases)
LIDAR(buildingmotions,pollution)
RADAR(building/streetvibrations,buildingmotion,trafficflow)

Correlativedataontheurbanscenes

Meteorology(temperature,winds,visibility)
Scenegeometry(distances,directions, identitiesoffeaturesvisible)
Parcelandlandusedata,buildingcharacteristicsandactivities,
buildingutilityconsumptions,andrealestatevaluationdata
Insitu pollutiondataandlocation/natureofmajorsources
Insitu vehicleandpedestriantrafficforthestreetsvisible
Demographicandeconomicdata

Capabilitytoarchive,process,andanalyzedataacquired

Imageprocessingchains
Datawarehouse,GIS,Visualizationtools
Softwareandprocedurestoenhanceprivacyprotection

Personnelandfundingtocreateandoperatetheabove

Hyperspectral Imaging of Manhattan Bridge Lights

Source:Dobler,etal.

59

PersistentLWIRImagingofManhattan
CUSP+AerospaceCorporation
April615fromHoboken
WestSidefromtheBatteryto~59Street

128spectralchannelscovering7.6 13.6m
PlumedetectionandmolecularID

61

Northernportionoftheviewfrom
Hoboken

GBSSset
up

Shortterm
variabilityinthe
thermalIR
Referenceimage

~1minlater

Difference
image

Shortterm
variabilityinthe
thermalIRdue
toacoolingtower
lightingup.

Temperature/EmissivitySeparation
R
T

Materials
Reflections

Broadband
Thermography
Forbuildingenvelops

Spectralanalysis
showsdiverse,
episodicplumes

ThermalImage

Ammonia

Difluoroethane (Freon)

AtypicalspectralfittopixelswithNH3
NH3 spectraltemplate
Datafitwithtemplate
Residualtofit

TStatisticforfittoeach
of
700compoundsinthe
library;
NH3 isahitwitht~10
8

CO2 at200C

Controled Release Diflouroethane


NorthofChelseapiers

Onecanateachofthreelocations

ThermalImage

Freon22appearsinabsorption
andemission
Ozonedestroyingmolecule
beingphasedoutintheUS

Chlorodifluoromethane (Freon22)

Chlorodifluoromethane (Freon22)

Chlorodifluoromethane (Freonne)

NumberofCapturesperMoleculein9days
ID

CountofID

Ammonia
Chlorodifluoromethane
CarbonDioxide(HITRAN200C)
Difluoromethane
CarbonDioxide(HITRAN300C)
1,1,1,2Tetrafluoroethane
Pentafluoroethane
CarbonDioxide(HITRAN100C)
Methane
Acetyliodide
Sulfurdioxide
Methane(HITRAN5C)
Chlorofluoromethane
2(Diisopropylamino)ethanol
Cyclohexanol
CarbonDioxide(HITRAN50C)
Othercomponents

9,816
1,722
1,051
412
211
197
190
162
83
65
57
39
36
29
29
28
416

TOTAL

14,543

AcetonePlume
April7,2015MidtownManhattan
(1.5kmdistance)4mx6m

CoutureCleaners
679WashingtonStreet

Acetoneisusedextensively
indrycleaning

Whatssuccessafter5years?
DefineandelaborateUrbanScience
Avibrantworldclasscenterpursuingsuch
NucleateanNYU/NYUPolycommunity
ImplementCUSPfacilities

ProjectsthatimpacttheCityanditsCitizens
CUSPestablishedasatrustedpartnertoNYC
Supportpublicunderstandingandengagement

Trainseveralhundredpeopleinthisnewfield
CommercializationofCUSPtechnologies
Bringnewtoolstothesocialsciences
Begintofranchisethebrandglobally

ThankYou
cusp.nyu.edu
NYUCUSP
@NYUCUSP