Page 1
#AnalyticsXC o p y r ig ht © 201 6, SAS In st i tute In c. A l l r ig hts r ese rve d.
Best Practices Using Analytics and Data for Good: Disaster Response Case StudyBrian KellyHead of Community Stabilization UnitInternational Organization for Migration
I-Sah HsiehGlobal Manager, Int’l DevelopmentSAS
Page 2
C o p y r ig ht © 201 3, SAS In st i tute In c. A l l r ig hts r ese rve d.
AGENDA BEST PRACTICES – DATA FOR GOOD
• Data4Good overview
• Role of Data
• Role of Analytics
• Case studies:
• Disaster Responder’s analytic journey
• ACTwithSAS – a crowdsourcing initiative
Page 3
C o p y r ig ht © 201 3, SAS In st i tute In c. A l l r ig hts r ese rve d.
Page 4
C o p y r ig ht © 201 3, SAS In st i tute In c. A l l r ig hts r ese rve d.
Page 5
C o p y r ig ht © 201 3, SAS In st i tute In c. A l l r ig hts r ese rve d.
NEW FRONTIER BIG DATA – GETS BIGGER!
Page 6
C o p y r ig ht © 201 3, SAS In st i tute In c. A l l r ig hts r ese rve d.
FIRST WAVE MACHINES TALKING TO PEOPLE
Page 7
C o p y r ig ht © 201 3, SAS In st i tute In c. A l l r ig hts r ese rve d.
Page 8
C o p y r ig ht © 201 3, SAS In st i tute In c. A l l r ig hts r ese rve d.
SECOND WAVE MACHINES TALKING TO MACHINES, THEN TO PEOPLE
Page 9
C o p y r ig ht © 201 3, SAS In st i tute In c. A l l r ig hts r ese rve d.
EXAMPLES “INTERNET OF THINGS ANALYTICS“
Connected Devices
Data
Insight
Action
Smart CitySmart Home Smart CarSmart Grid
Sensors embedded in
road surfaceConnected lock
Bluetooth dongle
reading vehicle data
Connected PMUs
Detect weather, predict
traffic, sense 911 vehiclesDetect when I leave
Predict unsafe
conditions & likely collision
Detect voltage instabilities
on transmission network
Adjust speed limits and
shift lane directions to optimize traffic flow
Lock the door, set the
alarm and adjust the HVAC
Slow the vehicle, alert
the driver
Trigger corrective
actions to maintain grid reliability
Page 10
C o p y r ig ht © 201 3, SAS In st i tute In c. A l l r ig hts r ese rve d.
BIG DATA ALTERNATE DEFINITION
Big Data = More data than what you currently use!
Page 11
C o p y r ig ht © 201 3, SAS In st i tute In c. A l l r ig hts r ese rve d.
DATA 4 GOOD BEST PRACTICES
• Where to begin?
• Answer the hard questions OR Let the DATA tell the story
•
• Use Analytics
• Visualization, Data mining, Forecasting, Predictive, Optimization
• Use more & new data sources (non-traditional data sources, BIG data)
• Consider new technologies (mobile devices, Cloud computing, etc.)
Page 12
C o p y r ig ht © 201 3, SAS In st i tute In c. A l l r ig hts r ese rve d.
ENGAGING MIGRATION &
DISPLACEMENT DATA
Page 13
C o p y r ig ht © 201 3, SAS In st i tute In c. A l l r ig hts r ese rve d.
INTERNATIONAL ORGANIZATION FOR MIGRATION (IOM)
1953
Page 14
C o p y r ig ht © 201 3, SAS In st i tute In c. A l l r ig hts r ese rve d.
INTERNATIONAL ORGANIZATION FOR MIGRATION (IOM)
2015
http://medialib.iom.int
Page 15
C o p y r ig ht © 201 3, SAS In st i tute In c. A l l r ig hts r ese rve d.
CASE STUDIES IOM’S ANALYTIC JOURNEY
Improving INSIGHT, HINDSIGHT, and FORESIGHT:
• Typhoon Haiyan (2013)
• Nepal Earthquake (2015)
• Mediterranean Migration Crisis (2015)
• Cyclone Winston (2016)
Page 16
C o p y r ig ht © 201 3, SAS In st i tute In c. A l l r ig hts r ese rve d.
DATA DISPLACEMENT TRACKING MATRIX
Page 17
C o p y r ig ht © 201 3, SAS In st i tute In c. A l l r ig hts r ese rve d.
VISUAL ANALYTICS “WHICH SITES HAVE MOST CRITICAL NEEDS?”
FASTER
INSIGHTS
Page 18
C o p y r ig ht © 201 3, SAS In st i tute In c. A l l r ig hts r ese rve d.
VISUAL ANALYTICS MOST COMMON PHRASES IN “HEALTH PROBLEMS”
FASTER
INSIGHTS
Elderly
Page 19
C o p y r ig ht © 201 3, SAS In st i tute In c. A l l r ig hts r ese rve d.
VISUAL ANALYTICS HOW DO WE GET PEOPLE HOME?
FASTER
INSIGHTS
Page 20
C o p y r ig ht © 201 3, SAS In st i tute In c. A l l r ig hts r ese rve d.
DEEPER HINDSIGHT LOOKING ACROSS 4 TYPHOONS IN PHILIPPINES
Page 21
C o p y r ig ht © 201 3, SAS In st i tute In c. A l l r ig hts r ese rve d.
VISUAL ANALYTICS PREDICTIVE ANALYTICS – GENDER BASED VIOLENCE
IMPROVE
FORESIGHT
Page 22
C o p y r ig ht © 201 3, SAS In st i tute In c. A l l r ig hts r ese rve d.
VISUAL ANALYTICS PREDICTIVE ANALYTICS – CAUSES OF DEATH (TEXT)
IMPROVE
FORESIGHT
Page 23
C o p y r ig ht © 201 3, SAS In st i tute In c. A l l r ig hts r ese rve d.
BIG DATA SOCIAL MEDIA & GLOBAL TRADE DATA
Page 24
C o p y r ig ht © 201 3, SAS In st i tute In c. A l l r ig hts r ese rve d.
CROWDSOURCING “ACT” WITH SAS - HUMANITARIAN ASSISTANCE
When disaster strikes…
• Time is critical
• Manpower is limited
• Analysis is often delayed
• External help is often unorganized
Page 25
C o p y r ig ht © 201 3, SAS In st i tute In c. A l l r ig hts r ese rve d.
CROWDSOURCING “ACT” WITH SAS - HUMANITARIAN ASSISTANCE
• What if we could mobilize thousands of people to:
• Review data?
• Answer questions using leading analytics tools?
• Bring new questions, data, and ideas?
• SAS provides the cloud-based analytic platform
• Help on their own schedule
• No travel required (only internet)
Page 26
C o p y r ig ht © 201 3, SAS In st i tute In c. A l l r ig hts r ese rve d.
CROWDSOURCING EXAMPLE: MISSING MIGRANTS PROGRAM
• In first half of 2016, migrant deaths increased 28%
• Why? What are the dangers migrants face by route? By region? By demographic?
• Small team (<3 people)
• Based in Berlin, Germany
Page 27
C o p y r ig ht © 201 3, SAS In st i tute In c. A l l r ig hts r ese rve d.
CROWDSOURCING DEMO
Page 28
C o p y r ig ht © 201 6, SAS In st i tute In c. A l l r ig hts r ese rve d.
#AnalyticsX