Data Science for Driving Urban Intelligence : On the Ground Case - - PowerPoint PPT Presentation

Data Science for Driving Urban Intelligence :

On the Ground Case Studies in Cities

Presented by Amen Ra Mashariki

What keeps City Leadership Up at Night?

Our world is changing...

Precipitation from heavy storms

70 %

since 1958

New Era of City Weather.

Frequency of 7.5t floods (~ 2 per millenium)

Every 5 yrs

By 2030

PREDICTED SEA LEVEL RISE 18 to 50 inches

By 2100

U.S. POPULATION LIVING IN URBAN AREAS -82%

64%

Increase since 1950

CONSTRUCTION ACTIVITY

4-6%

Annual growth rate

Urbanization.

Overview

$ 270B

Replacement cost for gas pipes.

Aging Infrastructure.

ELECTRIC INFRASTRUCTURE

$177B

Investment gap 2016 - 2025

Changing world.

Nearly half of the local Government workforce will be nearing retirement age.

Workforce Turnover.

RISK

External Threats Internal Vulnerabilities

Vegetation Contractors Pollution Weather and more...

Government Leaders Are Interested In Outcomes For Their Communities…

Healthy Economically Vibrant Safe Diverse Business Opportunities Minimizing traffic congestion and fatalities Sustainability

What Else?

Precision and Accuracy Cost- Effectiveness Timely Collaborative Transparent and Ethical Equitable With Data?

Citywide Analytics Offices (National / Global)

Glasgow, GCC - Jenny O’Hagan Worcestire Office

• f Data Analytics

(WODA), Neill Crump London Office of Data Analytics (LODA) – Theo Blackwell City Data Analytics Office , Barcelona Chicago San Francisco Los Angeles NYC KC Boston New Orleans

Strong Citywide Analytics Strategy

Urban Intelligence ECOSYSTEM

Analytics Patterns

Prioritizing Where to go first? Scenario Analysis What if? Anomaly Detection What is out of the

• rdinary?

Ranking a list according to certain criteria can enable more efficient use of resources. While the Department of Education must make all schools ADA-compliant, MODA used DOE data to prioritize which schools to renovate first in order to reduce the number of disabled students who needed to be bussed in the meantime. Useful when getting to the worst things earlier can mitigate negative effects. Considering alternative events and measuring the range of their possible outcomes can help define the best course of action. As part of the Mayor’s Office of Long Term Planning and Sustainability’s research for a new commercial composting policy, MODA predicted how much waste local businesses would generate under various regulatory

• thresholds. Useful for planning for a range of possible outcomes.

Some processes can be improved by identifying and investigating outliers. For example, registration records may have a number of files that display unusual

• characteristics. Flagging those records and examining them may reveal procedural
• versight or fraudulent transaction. Useful when investigating the

“out-of-the-

• rdinary” is more feasible than examining every case.

Estimating How much? Matching What goes with what? Targeting Where to look?

Matching can optimally pair two groups against a certain set of constraints. When the appointment scheduler for IDNYC was backlogged with duplicate requests, MODA helped match applicants to times and locations based on indicated

• preferences. Useful for equitably distributing a limited set of resources.

Projects can be planned more effectively when time, material, and costs are estimated in advance. MODA worked with the Department

• f

Housing Preservation and Development to estimate the resource requirements and program outcomes for a new set of “Enhanced Contractor Review” oversight

• procedures. Useful for quantifying the costs and benefits of new programs.

Targeting can narrow an operational domain to enable better resources allocation. MODA created a model to help identify buildings that have displayed a pattern of unsafe living conditions. This enabled the Tenant Harassment Prevention Task Force to follow up with inspections and enforcement actions when necessary. Useful for identifying a subset for a specific intervention.

Other “analytics” programs…

Performance Management

Approach

Evaluation of Operations Research Analytics Open Data

Define, visualize, often using dashboards, and manage to KPIs

Process

Assess a project, program or policy design or results Define and assess alternatives using a broad range of tools Publish civic data for use by the City and the public Meet goals and KPI targets

Outcome

Better investment of resources; Better policy decisions Report or memo with policy or program recommendations Drive transparency and public engagement

Is The Problem Solvable With Data?

What Should Your Team look Like?

Case Studies

Accuracy and expediency are crucial. Limited human and financial resources for inspection and canvassing. Do not want another person in your city to get sick from Legionella. Over a million buildings in your city.

Data Science and Location Intelligence Can Help…

Without Data Science

• There was a 1 in 10 chance of us identifying a building that had a cooling

tower.

With Data Science

• There was an 8 in 10 chance that we could correctly identify a building

that had a cooling tower.

FIND THE COOLING TOWERS USING

THE SCIENCE OF WHERE

CITYWIDE OUTREACH EFFORT

Cooling Towers

Where are Commercial Cooling Tower Systems in New York City?

1 2 3 4 5 6 7 8 9 10 Baseline Logistic Intuition Variable Choice Stepwise VIF-d no Bldg Cat FLAM

• Avg. Model

Success ( out

• f 10)

"No Analytics" Comparison

ANSWER(S):

A machine learning model can correctly identify a building with a cooling tower 8

• ut of 10 times, compared to

1 out of 10 times without using analytics.

NYC Current Situation

Rent Regulation Laws and Deregulation Loopholes High Market Rate Rents

Tenant Harassment

High-Profile Arrests

BBL: 3014000042, BIN: 3037671 Owner: Daniel Melamed BBL: 3033320008, BIN: 3076261 Owner: Israel Brothers

1578 Union St. 98 Linden St.

Which Are Predictive? How Can We Test?

• Building has rent stabilized

apartments

• Large price difference between

market rate and rent stab.

• Recent sale of the buildings TEST
• Recent sale of the building

followed by construction TEST

• 311 Complaints TEST

Use The Loss Of Rent Stabilized Apartments As A Proxy Likely Conditions

Recently Sold Buildings

Possible Indicators Of Tenant Harassment In City Data

ANALYSIS QUESTION

Which of these indicators is more likely to result in the loss of rent regulated units?

Recently Sold Building Followed By Intense Construction Illegal Construction Complaints from residents Large Difference Between Market Rate and Rent Regulated Apartments Landlords Taking Tenants To Court

Which data elements and sources are available for MODA to explore?

DOF CHARGES DATA

• Rent Stabilization
• HPD Emergency Charges

DOF OWNERSHIP

• Sales / Change in Ownership
• Names / Addresses of Owners

311 SERVICE REQUESTS

• HPD Related
• Construction Related
• Categories of Complaint Types

JOHN KRAUSS DATA

• Rent Stab unit counts
• Pulled from PDF tax bills
• Has counts for each RS BBL from 2007 - 2014

BBL

DOB CONSTRUCTION JOBS

• Pre-filing Dates
• Construction Type

DATA SOURCES

Coordinated GIS Data Integration

Apps Desktop APIs

In 2014 about 5,000 buildings experienced unit loss. 3,000 of them lost 1 unit, 1000 lost 2 units.

Number of BBLs Loosing RS units, colors indicate numbers of units lost, blue=1, purple=2, …

Around 5k BBLs (over 10%) experienced unit loss from 2013-2014

Identifying Trends: Rent Stab Unit Loss

• DOB Illegal Work Complaints

Buildings with dust or asbestos complaints were 4-7 times more likely to experience unit loss in the following year compared to a random rent stabilized building.

• Sales Followed By Construction
• 311 SR Related To Construction
• 311 SR For Air Quality Non-Construction

What We Found

38