Twitter: @NCD4G DATA SCIENCE FOR SOCIAL GOOD NCData4Good.github.io
Zeydy Ortiz, Ph. D.
Organizer
Agenda
What is data science?
How can data science help social good
organizations?
Better understand people served
Design more efficient programs
Better allocate resources
What is NC Data4Good?
Twitter: @NCD4G DATA SCIENCE FOR SOCIAL GOOD NCData4Good.github.io
Twitter: @NCD4G DATA SCIENCE FOR SOCIAL GOOD NCData4Good.github.io
Data
Scientist
Zeydy Ortiz, Ph. D.
zortiz@datacrunchlab.com
Twitter: @DrZeydy | @DCrunchLab
Computer Engineer
Computer Science
Performance
Engineer
Data journey to value
Twitter: @NCD4G DATA SCIENCE FOR SOCIAL GOOD NCData4Good.github.io
DATA VALUE
GENERATION
COLLECTION
AGGREGATION
RESULTS
What happened?
What is likely to happen?
What is the next best action?
ACTION
INTERPRETATION
DECISION
PRESCRIPTIVE
PREDICTIVE
DESCRIPTIVE
How can data
science help social
good organizations?
Better understand people served
Design more efficient programs
Better allocate resources
Twitter: @NCD4G DATA SCIENCE FOR SOCIAL GOOD NCData4Good.github.io
Sharing data to understand homelessness
DataKind UK for St. Mungo’s Broadway & Citizens Advice
http://www.datakind.org/projects/sharing-data-to-learn-about-homelessness
Twitter: @NCD4G DATA SCIENCE FOR SOCIAL GOOD NCData4Good.github.io
BETTER UNDERSTAND PEOPLE SERVED
Understanding needs
The St Mungo’s Broadway team
sought to answer questions such
as:
•What sorts of advice do people
seek before they become
homeless?
•How do St Mungo’s Broadway’s
services perform for different
groups of people?
•What happens to people when
they leave St Mungo’s Broadway’s
services?
Twitter: @NCD4G DATA SCIENCE FOR SOCIAL GOOD NCData4Good.github.io
Identify “superusers”
DataKind & Pivotal for Crisis Text Line
http://www.fastcoexist.com/3047638/how-data-science-shaped-this-teen-counseling-by-text-service
Twitter: @NCD4G DATA SCIENCE FOR SOCIAL GOOD NCData4Good.github.io
DESIGN MORE EFFICIENT PROGRAM
Designing more efficient
program
Situation: 3% texters were
superusers using 34% of
counselor’s conversation minutes
(40-50 minutes per conversation)
Result: Identified superusers as
early as their 5th conversation
=> using only 8% of conversation
minutes
Twitter: @NCD4G DATA SCIENCE FOR SOCIAL GOOD NCData4Good.github.io
Prioritizing hazardous waste inspections
Data Science for Social Good Fellowship for Environmental Protection Agency (EPA)
https://dssg.uchicago.edu/2015/10/16/epa-data-driven-hazardous-waste-detection/
Twitter: @NCD4G DATA SCIENCE FOR SOCIAL GOOD NCData4Good.github.io
BETTER ALLOCATE RESOURCES
Better allocate resources
Twitter: @NCD4G DATA SCIENCE FOR SOCIAL GOOD NCData4Good.github.io
Situation: In 2014, there were
more than 20,000 chemical spills
in the US leading to more than 800
deaths and $50 million in property
damage
Challenge: Prioritize facilities to
inspect (~4% of active facilities)
Results: Risk score model to
effectively identify potential
violators, better allocate
inspection resources, and
maximize the impact of each
investigation
Other Data4Good projects
Twitter: @NCD4G DATA SCIENCE FOR SOCIAL GOOD NCData4Good.github.io
Twitter: @NCD4G DATA SCIENCE FOR SOCIAL GOOD NCData4Good.github.io
Twitter: @NCD4G DATA SCIENCE FOR SOCIAL GOOD NCData4Good.github.io
Making an
Impact
Data
Scientists
Supporters
Bring together social
good organizations
and data scientists to
make a difference in
our communities
Social Good
Organizations
COMMUNITY
Twitter: @NCD4G DATA SCIENCE FOR SOCIAL GOOD NCData4Good.github.io
Data
Scientists
Supporters
DataCrunch
2015
Enable social
innovators to
address issues of
childhood hunger
and access to fresh,
nutritious foods
Social Good
Organizations
COMMUNITY
Twitter: @NCD4G DATA SCIENCE FOR SOCIAL GOOD NCData4Good.github.io
Twitter: @NCD4G DATA SCIENCE FOR SOCIAL GOOD NCData4Good.github.io
1 in 5 children living in poverty
Darker areas represent
~1000 children in
poverty:
• North Durham
• East Durham city
• North (east) Raleigh
• South East Raleigh
• Near US401S &
TenTen
• Garner
• Benson
Where is the need?
Twitter: @NCD4G DATA SCIENCE FOR SOCIAL GOOD NCData4Good.github.io
Food pantries near bus stops
% Pantries within 1 mile
of a bus stop:
• Durham : 95%
• Orange : 67%
• Wake : 50%
• Johnston: 0%
Where are the gaps?
Many food pantries are not
accessible by public
transportation
Twitter: @NCD4G DATA SCIENCE FOR SOCIAL GOOD NCData4Good.github.io
Twitter: @NCD4G DATA SCIENCE FOR SOCIAL GOOD NCData4Good.github.io
Thanks to the community!
“I learned a lot about both the content (childhood hunger in the community) and techniques,
got to apply my skills and contribute to the project, and made new friends”
Twitter: @NCD4G DATA SCIENCE FOR SOCIAL GOOD NCData4Good.github.io
Thank you!
Zeydy Ortiz, Ph. D.
zortiz@datacrunchlab.com
Twitter: @NCD4G DATA SCIENCE FOR SOCIAL GOOD NCData4Good.github.io

Data Science for Social Good

  • 1.
    Twitter: @NCD4G DATASCIENCE FOR SOCIAL GOOD NCData4Good.github.io Zeydy Ortiz, Ph. D. Organizer
  • 2.
    Agenda What is datascience? How can data science help social good organizations? Better understand people served Design more efficient programs Better allocate resources What is NC Data4Good? Twitter: @NCD4G DATA SCIENCE FOR SOCIAL GOOD NCData4Good.github.io
  • 3.
    Twitter: @NCD4G DATASCIENCE FOR SOCIAL GOOD NCData4Good.github.io Data Scientist Zeydy Ortiz, Ph. D. zortiz@datacrunchlab.com Twitter: @DrZeydy | @DCrunchLab Computer Engineer Computer Science Performance Engineer
  • 4.
    Data journey tovalue Twitter: @NCD4G DATA SCIENCE FOR SOCIAL GOOD NCData4Good.github.io DATA VALUE GENERATION COLLECTION AGGREGATION RESULTS What happened? What is likely to happen? What is the next best action? ACTION INTERPRETATION DECISION PRESCRIPTIVE PREDICTIVE DESCRIPTIVE
  • 5.
    How can data sciencehelp social good organizations? Better understand people served Design more efficient programs Better allocate resources Twitter: @NCD4G DATA SCIENCE FOR SOCIAL GOOD NCData4Good.github.io
  • 6.
    Sharing data tounderstand homelessness DataKind UK for St. Mungo’s Broadway & Citizens Advice http://www.datakind.org/projects/sharing-data-to-learn-about-homelessness Twitter: @NCD4G DATA SCIENCE FOR SOCIAL GOOD NCData4Good.github.io BETTER UNDERSTAND PEOPLE SERVED
  • 7.
    Understanding needs The StMungo’s Broadway team sought to answer questions such as: •What sorts of advice do people seek before they become homeless? •How do St Mungo’s Broadway’s services perform for different groups of people? •What happens to people when they leave St Mungo’s Broadway’s services? Twitter: @NCD4G DATA SCIENCE FOR SOCIAL GOOD NCData4Good.github.io
  • 8.
    Identify “superusers” DataKind &Pivotal for Crisis Text Line http://www.fastcoexist.com/3047638/how-data-science-shaped-this-teen-counseling-by-text-service Twitter: @NCD4G DATA SCIENCE FOR SOCIAL GOOD NCData4Good.github.io DESIGN MORE EFFICIENT PROGRAM
  • 9.
    Designing more efficient program Situation:3% texters were superusers using 34% of counselor’s conversation minutes (40-50 minutes per conversation) Result: Identified superusers as early as their 5th conversation => using only 8% of conversation minutes Twitter: @NCD4G DATA SCIENCE FOR SOCIAL GOOD NCData4Good.github.io
  • 10.
    Prioritizing hazardous wasteinspections Data Science for Social Good Fellowship for Environmental Protection Agency (EPA) https://dssg.uchicago.edu/2015/10/16/epa-data-driven-hazardous-waste-detection/ Twitter: @NCD4G DATA SCIENCE FOR SOCIAL GOOD NCData4Good.github.io BETTER ALLOCATE RESOURCES
  • 11.
    Better allocate resources Twitter:@NCD4G DATA SCIENCE FOR SOCIAL GOOD NCData4Good.github.io Situation: In 2014, there were more than 20,000 chemical spills in the US leading to more than 800 deaths and $50 million in property damage Challenge: Prioritize facilities to inspect (~4% of active facilities) Results: Risk score model to effectively identify potential violators, better allocate inspection resources, and maximize the impact of each investigation
  • 12.
    Other Data4Good projects Twitter:@NCD4G DATA SCIENCE FOR SOCIAL GOOD NCData4Good.github.io
  • 13.
    Twitter: @NCD4G DATASCIENCE FOR SOCIAL GOOD NCData4Good.github.io
  • 14.
    Twitter: @NCD4G DATASCIENCE FOR SOCIAL GOOD NCData4Good.github.io
  • 15.
    Making an Impact Data Scientists Supporters Bring togethersocial good organizations and data scientists to make a difference in our communities Social Good Organizations COMMUNITY Twitter: @NCD4G DATA SCIENCE FOR SOCIAL GOOD NCData4Good.github.io
  • 16.
    Data Scientists Supporters DataCrunch 2015 Enable social innovators to addressissues of childhood hunger and access to fresh, nutritious foods Social Good Organizations COMMUNITY Twitter: @NCD4G DATA SCIENCE FOR SOCIAL GOOD NCData4Good.github.io
  • 17.
    Twitter: @NCD4G DATASCIENCE FOR SOCIAL GOOD NCData4Good.github.io
  • 18.
    1 in 5children living in poverty Darker areas represent ~1000 children in poverty: • North Durham • East Durham city • North (east) Raleigh • South East Raleigh • Near US401S & TenTen • Garner • Benson Where is the need? Twitter: @NCD4G DATA SCIENCE FOR SOCIAL GOOD NCData4Good.github.io
  • 19.
    Food pantries nearbus stops % Pantries within 1 mile of a bus stop: • Durham : 95% • Orange : 67% • Wake : 50% • Johnston: 0% Where are the gaps? Many food pantries are not accessible by public transportation Twitter: @NCD4G DATA SCIENCE FOR SOCIAL GOOD NCData4Good.github.io
  • 20.
    Twitter: @NCD4G DATASCIENCE FOR SOCIAL GOOD NCData4Good.github.io
  • 21.
    Thanks to thecommunity! “I learned a lot about both the content (childhood hunger in the community) and techniques, got to apply my skills and contribute to the project, and made new friends” Twitter: @NCD4G DATA SCIENCE FOR SOCIAL GOOD NCData4Good.github.io
  • 22.
    Thank you! Zeydy Ortiz,Ph. D. zortiz@datacrunchlab.com Twitter: @NCD4G DATA SCIENCE FOR SOCIAL GOOD NCData4Good.github.io

Editor's Notes

  • #2 How we want to bring together data science professionals to make an impact in our communities, using their skills for social good help mission-driven organizations through programs to identify relevant data to better understand the people they serve; bring out insights to design more efficient programs; and anticipate future needs to better allocate resources
  • #14 How we want to bring together data science professionals to make an impact in our communities, using their skills for social good help mission-driven organizations through programs to identify relevant data to better understand the people they serve; bring out insights to design more efficient programs; and anticipate future needs to better allocate resources