View
0
Download
0
Category
Preview:
Citation preview
사물인터넷 환경에서 비즈니스 인사이트 중심의데이터 분석 전략
한국 IBM진승의 실장jse@kr.ibm.com
1.1. 사물인터넷으로사물인터넷으로 인한인한 데이터데이터 분석분석 및및 관리관리 환경의환경의 변화변화
2.2. Actionable InsightsActionable Insights의의 개념개념 및및 중요성중요성
3.3. Actionable Insights Actionable Insights 도출을도출을 위한위한 분석분석 고도화고도화 방안방안
4.4. 사물인터넷사물인터넷 환경에서의환경에서의 Actionable Insights Actionable Insights 기반기반 데이터데이터 분석분석 사례사례 연구연구
AgendaAgenda
© Copyright International Business Machines Corporation 2014, All rights reserved 1
1.1. 사물인터넷으로사물인터넷으로 인한인한 데이터데이터 분석분석 및및 관리관리 환경의환경의 변화변화
2.2. Actionable InsightsActionable Insights의의 개념개념 및및 중요성중요성
3.3. Actionable Insights Actionable Insights 도출을도출을 위한위한 분석분석 고도화고도화 방안방안
4.4. 사물인터넷사물인터넷 환경에서의환경에서의 Actionable Insights Actionable Insights 기반기반 데이터데이터 분석분석 사례사례 연구연구
The Internet of Things
Definition1
The Internet of Things refers to the use of sensors, actuators, and data communications technology built into physical objects - from roadways to pacemakers - that enable those objects to be tracked, coordinated, or controlled across a data network or the InternetThere are three steps in Internet of Things applications:§Capturing data from the object (for example,
simple location data or more complex information),
§Aggregating that information across a data network, and
§Acting on that information - taking immediate action or collecting data over time to design process improvements.
© Copyright International Business Machines Corporation 2014, All rights reserved 2
Source: 1. Disruptive Technologies, McKinsey Global Institute, May 2013
Definition1
The Internet of Things refers to the use of sensors, actuators, and data communications technology built into physical objects - from roadways to pacemakers - that enable those objects to be tracked, coordinated, or controlled across a data network or the InternetThere are three steps in Internet of Things applications:§Capturing data from the object (for example,
simple location data or more complex information),
§Aggregating that information across a data network, and
§Acting on that information - taking immediate action or collecting data over time to design process improvements.
Internet of Things continues to explode
© Copyright International Business Machines Corporation 2014, All rights reserved 3
Source: ABI Research, Cisco ‘The Internet of Things: How the Next Evolution of the Internet is Changing Everything’, Cisco Visual Networking Index (VNI) Global Mobile Data Traffic Forecast for 2010-2015, IBM Analysis
Internet of Things Opportunity
50BDevices20201> Compute Economics
+ Ubiquitous Connectivity+ Big Data Analytics
Today 85% of deployed systems are unconnected, do not share data with each other or the cloud.
And new devices are being added every day.
© Copyright International Business Machines Corporation 2014, All rights reserved 4
= BusinessTransformation
15BDevices2015
2BDevices
2006
1IDC*, Intel, United Nations3McKinsey Global Institute*
2IDC Digital Universe Study, Dec 2012*Other names and brands may be claimed as the property of others.
> Compute Economics+ Ubiquitous Connectivity+ Big Data Analytics
1 Connect2 Collect Data3 Analyze4 Transform
GSMA “Connected Life” forecast $4.5T in 2020
Connected Life is everything that is connected and how they interact: cars, mobile devices, buildings, sensors and people
Top Ten in 20201. Connected Car $600 billion2. Clinical Remote Monitoring $350 billion3. Assisted Living $270 billion4. Home and Building Security $250 billion5. Pay-As-You-Drive Car Insurance $245 billion6. New Business Models for Car Usage $225 billion7. Smart Meters $105 billion8. Traffic Management $100 billion9. Electric Vehicle Charging $75 billion10.Building Automation $40 billion
© Copyright International Business Machines Corporation 2014, All rights reserved 5
Connected Life is everything that is connected and how they interact: cars, mobile devices, buildings, sensors and people
Top Ten in 20201. Connected Car $600 billion2. Clinical Remote Monitoring $350 billion3. Assisted Living $270 billion4. Home and Building Security $250 billion5. Pay-As-You-Drive Car Insurance $245 billion6. New Business Models for Car Usage $225 billion7. Smart Meters $105 billion8. Traffic Management $100 billion9. Electric Vehicle Charging $75 billion10.Building Automation $40 billion
Source:http://www.globaltelecomsbusiness.com/article/2985699/Connected-devices-will-be-worth-45t.html
Internet of Things is turning Big Data into Massive Data
© Copyright International Business Machines Corporation 2014, All rights reserved 6
IoT use cases have many common requirements
Core Requirements:§ Easily on-board connected “things”§ Create a real-time communication channel with the “thing”§ Begin capturing data from the “thing”§ Visualize data from the “thing”§ Collect data in a historian DB§ Provide access to the collected data§ Manage the “things” and the connectivity to them§ Secure the data from the “thing” and control access to that that data§ Pay for the service based on usage
Extended Requirements:§ Perform analytics both in real-time and on historical trend data§ Trigger events based on specific data conditions§ Interact with the “thing” from business apps and/or from mobile devices§ Send commands to the “thing”
© Copyright International Business Machines Corporation 2014, All rights reserved 7
Core Requirements:§ Easily on-board connected “things”§ Create a real-time communication channel with the “thing”§ Begin capturing data from the “thing”§ Visualize data from the “thing”§ Collect data in a historian DB§ Provide access to the collected data§ Manage the “things” and the connectivity to them§ Secure the data from the “thing” and control access to that that data§ Pay for the service based on usage
Extended Requirements:§ Perform analytics both in real-time and on historical trend data§ Trigger events based on specific data conditions§ Interact with the “thing” from business apps and/or from mobile devices§ Send commands to the “thing”
Brand New Trends of Information Management
Back office, business transaction systems- Relational database- Relation of the entities described as relational tables- Rigid data schema
Mobile, social, Internet-of-People- Graph database, JSON store- Relation of the entities described as graphs and events/objects- Flexible data schema
© Copyright International Business Machines Corporation 2014, All rights reserved 8
Internet-of-Things, physical world- Relation among data entities follows physical model- Ideally the information management system should capture both the data entities and their relationship- Current solution: separate data and relationship (model), using RDB or file to store the data, leave model & analytics to applications
Imagine the Possibilities of Analyzing All this Data in Real-time
© Copyright International Business Machines Corporation 2014, All rights reserved 9
Predictive Maintenance & Quality
§ Estimate and extend component life
§ Increase return on assets
§ Improve maintenance, inventory and resource schedules
§ Improve quality and reduce recalls
§ Reduce time to identify issues
§ Improve readiness and service
ReduceReduceOperational costsOperational costs
ImproveImproveAsset productivityAsset productivity
© Copyright International Business Machines Corporation 2014, All rights reserved 10
§ Estimate and extend component life
§ Increase return on assets
§ Improve maintenance, inventory and resource schedules
§ Improve quality and reduce recalls
§ Reduce time to identify issues
§ Improve readiness and service
ImproveImproveAsset productivityAsset productivity
IncreaseIncreaseProcess efficiencyProcess efficiency
Predictive Maintenance & Quality (cont’d)
§ Monitor, maintain and optimize assets for better availability, utilization and performance
§ Predict asset failure to optimize product quality and supply chain processes
§ Remove guesswork from the decision-making process
© Copyright International Business Machines Corporation 2014, All rights reserved 11
§ Monitor, maintain and optimize assets for better availability, utilization and performance
§ Predict asset failure to optimize product quality and supply chain processes
§ Remove guesswork from the decision-making process
Combined with out-of-box models, dashboards, reports and source connectors
1.1. 사물인터넷으로사물인터넷으로 인한인한 데이터데이터 분석분석 및및 관리관리 환경의환경의 변화변화
2.2. Actionable InsightsActionable Insights의의 개념개념 및및 중요성중요성
3.3. Actionable Insights Actionable Insights 도출을도출을 위한위한 분석분석 고도화고도화 방안방안
4.4. 사물인터넷사물인터넷 환경에서의환경에서의 Actionable Insights Actionable Insights 기반기반 데이터데이터 분석분석 사례사례 연구연구
AgendaAgenda
© Copyright International Business Machines Corporation 2014, All rights reserved 12
1.1. 사물인터넷으로사물인터넷으로 인한인한 데이터데이터 분석분석 및및 관리관리 환경의환경의 변화변화
2.2. Actionable InsightsActionable Insights의의 개념개념 및및 중요성중요성
3.3. Actionable Insights Actionable Insights 도출을도출을 위한위한 분석분석 고도화고도화 방안방안
4.4. 사물인터넷사물인터넷 환경에서의환경에서의 Actionable Insights Actionable Insights 기반기반 데이터데이터 분석분석 사례사례 연구연구
Information Supply Chain of Internet of Things
© Copyright International Business Machines Corporation 2014, All rights reserved 13
Extracting Actionable Insights from Internet of Things
OPPORTUNITY:
We can realize the Smarter Planet vision by improving the intelligentpart
§ We can radically improve our ability to manage, control and optimize large-scale physical infrastructures
§ The business opportunities would be broad, applying to smarter buildings & cities, asset-intensive industries such as E&U, Oil&Gas, Automotive, etc.
§ The total value would be several $B for IBM software and services.
§ Profusion of sensor data and data sets from cities, buildings and homes
§ Scientific exploration: large open data sets for astronomy, meteorology, genome, etc.
§ More data collected from industry assets, e.g. for condition-based maintenance
TREND:
The physical world is becoming increasingly instrumented and interconnected
© Copyright International Business Machines Corporation 2014, All rights reserved 14
OPPORTUNITY:
We can realize the Smarter Planet vision by improving the intelligentpart
§ We can radically improve our ability to manage, control and optimize large-scale physical infrastructures
§ The business opportunities would be broad, applying to smarter buildings & cities, asset-intensive industries such as E&U, Oil&Gas, Automotive, etc.
§ The total value would be several $B for IBM software and services.
§ Transport & ingest unprecedented volumes of data from physical devices
§ New data stores and query languages for spatio-temporal and linked data
§ Cope with open world of data models; match data and analytics semantics and ontologies
§ Semi-automate extraction of statistical physics models
§ Real-time visualization & computational steering of what-if models
CHALLENGE:
We must intelligently manage & analyze big data from the physical world
Real-time Big Data Analytics of Internet of Things
Millions of events per second
Microsecond Latency
Real time insights
PowerfulAnalyticsAlgorithmic
TradingTelco ChurnPrediction
SmartGrid
CyberSecurity Government /
Law enforcement
ICUMonitoring
EnvironmentMonitoring
© Copyright International Business Machines Corporation 2014, All rights reserved 15
Millions of events per second
Microsecond Latency
IBM MessageSight
IBM InfoSphere Streams
Identify and leverage data needed for actionable insights
With new analytics, clients can better able to enter the world of predicting the next best offer, action or need
Demographicdata
Transactiondata
Pay Centers
Outbound calls
Call Centers
Events
Direct Mail
Kiosks
TransactionsOrders
Paymenthistory
Usage historyCharacteristics
Demographics
AttributesUtilities are capturing and leveraging internal and external data
© Copyright International Business Machines Corporation 2014, All rights reserved 16
Clients are working to understand how predictive insights about customers must be in order to succeed
With new analytics, clients can better able to enter the world of predicting the next best offer, action or need
Descriptive analyticsPredictive analyticsPrescriptive analytics
Interactiondata
Behavioraldata
Outbound calls
Website
Search
Online Advertising
MobileEmails
SMS/MMS
Social Media
Customer Service
Transaction
stage
E-mail / Chat
Call center notes
Web click-streamsIn-person
dialogs
Opinions
Preferences
Desires
Needs
Sample Actionable Insights Use Cases
Customer Targeting and Personalization
Customer Program Targeting
Customer Survey Selection
C&I Customer Experience and Engagement
Customer Recruitment
Program Eligibility Tracking
Targeted Marketing Campaigns
Business Capability Use Case Name
© Copyright International Business Machines Corporation 2014, All rights reserved 17
Customer Behavior Analysis
Identify Right Products and Services for Right Customer via Right Channel
Monitor Channel Preferences and Measure Channel Effectiveness
Monitor Transactions by Channel
Improve Targeting by Leveraging Customer Interaction and Usage
Correlate Channel Usage to Stated Preferences
Pay Arrangement and Credit
Branch Office Performance
Design Targeted Customer Offerings
Personalized Customer Service Via Customer Lifecycle Knowledge
Customer Engagement Initiative
1.1. 사물인터넷으로사물인터넷으로 인한인한 데이터데이터 분석분석 및및 관리관리 환경의환경의 변화변화
2.2. Actionable InsightsActionable Insights의의 개념개념 및및 중요성중요성
3.3. Actionable Insights Actionable Insights 도출을도출을 위한위한 분석분석 고도화고도화 방안방안
4.4. 사물인터넷사물인터넷 환경에서의환경에서의 Actionable Insights Actionable Insights 기반기반 데이터데이터 분석분석 사례사례 연구연구
AgendaAgenda
© Copyright International Business Machines Corporation 2014, All rights reserved 18
1.1. 사물인터넷으로사물인터넷으로 인한인한 데이터데이터 분석분석 및및 관리관리 환경의환경의 변화변화
2.2. Actionable InsightsActionable Insights의의 개념개념 및및 중요성중요성
3.3. Actionable Insights Actionable Insights 도출을도출을 위한위한 분석분석 고도화고도화 방안방안
4.4. 사물인터넷사물인터넷 환경에서의환경에서의 Actionable Insights Actionable Insights 기반기반 데이터데이터 분석분석 사례사례 연구연구
Managing IoT data using Timeseries
IBM Informix
Timeseries VTI Tables
IoT - Devices
Data Loader
JSONData Files
Middlew
are P
rocessing
JSON
© Copyright International Business Machines Corporation 2014, All rights reserved 19
Timeseries Tables
Timeseries VTI Tables
Data Loader
Middlew
are P
rocessing
type id Usage Timeseries(IFXTSBSON)
“XA” 12 (2014-01-01 01:21:000, {x:1,y:2}), (2014-02-02 01:23:000, {x:3, y:5, z:42})
“XB” 48 (2014-01-01 01:21:000, {c:1,d:”ACND”}), (2014-04-02 01:23:000, {c:92,d:”MCBS”, e:42})
“XC” 23 (2015-01-01 01:21:000, {p:1,q:2}), (2015-03-02 01:23:000, {p:3, y:5, z:42}),
Data
IoT - Devices
NoSQL used for IoT data
SQL {NoSQL:JSON}
Define Schema first Write the program first
Relational Key-value, Document, column family, graph and text
Changing schema is hard Assumes dynamic schema
© Copyright International Business Machines Corporation 2014, All rights reserved 20
Scale-up Scale-out
ACID consistency BASE consistency
Transactions No Transactions
SQL Proprietary API; Sometimes has the “spirit” of SQL
Timeseries also required for better IoT data analysis
SQL Timeseries
Define Schema first Create Timeseries Row Type
Relational Timeseries Optimized with projection to relational;
Often used with Spatial data
Changing schema is hard Changing schema is hard; Flexible with Timeseries({JSON})
Scale-up Scale-up & Scale-out
© Copyright International Business Machines Corporation 2014, All rights reserved 21
Scale-up Scale-up & Scale-out
ACID consistency ACID consistency
SQL SQL extensions; Relational projection
Timeseries also required for better IoT data analysis (cont’d)
• Instrument-generates large time-based Big Data– Data are time-based and time-serialized
– Stock trading, smart meters, network devices,
heavy industrial sensors. Etc.
• Characteristics of time-series data– Data ordered by time such as 15 minutes per read
– Data value varies based on time dimension
© Copyright International Business Machines Corporation 2014, All rights reserved 22
• Instrument-generates large time-based Big Data– Data are time-based and time-serialized
– Stock trading, smart meters, network devices,
heavy industrial sensors. Etc.
• Characteristics of time-series data– Data ordered by time such as 15 minutes per read
– Data value varies based on time dimension
ID, Values , Kwh
ID102,219,0.6
ID, Values , Kwh
ID102,220,0.6
ID, Values , Kwh
ID102,222,0.8
ID, Values , Kwh
ID102,219,0.7
01:00:00 01:15:00 01:30:00 01:45:00 Timeline
Examples of time-series data
ID101,220,0.5 ID101,220,0.8 ID101,215,0.6 ID101,218,0.5
Predictive Inspection Analytics
Improve Inspection Efficiency and Effectiveness: ØSupport Inspection with Data Driven Selection
• Examine all data attributes and rank them on importance to predicting successful investigations
ØPredictive Models and Advance Analytics• Find additional data attributes and predictive rule sets
that aren’t currently considering when selecting subjects for inspection
ØDecision Support and Integration• Recommended next best action in a complex
environment with different criteria and multiple inspection levels
© Copyright International Business Machines Corporation 2014, All rights reserved 23
Improve Inspection Efficiency and Effectiveness: ØSupport Inspection with Data Driven Selection
• Examine all data attributes and rank them on importance to predicting successful investigations
ØPredictive Models and Advance Analytics• Find additional data attributes and predictive rule sets
that aren’t currently considering when selecting subjects for inspection
ØDecision Support and Integration• Recommended next best action in a complex
environment with different criteria and multiple inspection levels
Predictive Inspection Analytics (cont’d)
Decision Management is a business discipline that applies advanced analytics to prioritize and optimize inspection selections.
§ Business rules to automate what you know
§ Models to predict what you don’t§ Optimization to make best use of
scarce resources§ Business events to identify
situations where action is needed
© Copyright International Business Machines Corporation 2014, All rights reserved 24
Decision Management is a business discipline that applies advanced analytics to prioritize and optimize inspection selections.
§ Business rules to automate what you know
§ Models to predict what you don’t§ Optimization to make best use of
scarce resources§ Business events to identify
situations where action is needed
IoT Mobility Analytics
© Copyright International Business Machines Corporation 2014, All rights reserved 25
1.1. 사물인터넷으로사물인터넷으로 인한인한 데이터데이터 분석분석 및및 관리관리 환경의환경의 변화변화
2.2. Actionable InsightsActionable Insights의의 개념개념 및및 중요성중요성
3.3. Actionable Insights Actionable Insights 도출을도출을 위한위한 분석분석 고도화고도화 방안방안
4.4. 사물인터넷사물인터넷 환경에서의환경에서의 Actionable Insights Actionable Insights 기반기반 데이터데이터 분석분석 사례사례 연구연구
AgendaAgenda
© Copyright International Business Machines Corporation 2014, All rights reserved 26
1.1. 사물인터넷으로사물인터넷으로 인한인한 데이터데이터 분석분석 및및 관리관리 환경의환경의 변화변화
2.2. Actionable InsightsActionable Insights의의 개념개념 및및 중요성중요성
3.3. Actionable Insights Actionable Insights 도출을도출을 위한위한 분석분석 고도화고도화 방안방안
4.4. 사물인터넷사물인터넷 환경에서의환경에서의 Actionable Insights Actionable Insights 기반기반 데이터데이터 분석분석 사례사례 연구연구
Well field optimization for Oil/Gas field service operations
§ Improve Production (reduce downtime or non-productive time) at lower costs AND
§Optimize the Safety Case (reduce operational risk)
§Operators must have both social and legal license to operate
§ Increasing reliance upon communications network and operational technology
Business Challenges
§Well head optimization§Public / private stakeholder visibility§Operational efficiency – remote monitoring with
pre-planned response plans§Reduce time to production and operational delays
thru more effective collaboration between operators, contractors, and public stakeholders
Opportunities for Innovation
© Copyright International Business Machines Corporation 2014, All rights reserved 27
§ Improve Production (reduce downtime or non-productive time) at lower costs AND
§Optimize the Safety Case (reduce operational risk)
§Operators must have both social and legal license to operate
§ Increasing reliance upon communications network and operational technology
Smar
t Dev
ices
What happens to the ‘well field optimization’ process if:§ The metering equipment fails? The communication network(s) fails? The application
or database servers have a performance issue ? Network security is breached?
Increasing supply chain efficiency from ‘Pit to Port’
§Decrease time to operation for new sites and improve return on capital investments
§ Improve production reliability (across supply chain)
§Operational compliance with environmental, health, and safety regulations
§ Increasing operational efficiency with real time visibility and process automation
Business Challenges
§Predict and act on potential failures§ Increasing production thru resource optimization§Operational efficiency – remote monitoring with
pre-planned response plans (internal / 3rd party)§Reduce time to production and operational delays
thru more effective collaboration between operators, contractors, and stakeholders
Opportunities for Innovation
© Copyright International Business Machines Corporation 2014, All rights reserved 28
§Decrease time to operation for new sites and improve return on capital investments
§ Improve production reliability (across supply chain)
§Operational compliance with environmental, health, and safety regulations
§ Increasing operational efficiency with real time visibility and process automation
Managing complexity of connected cars to reduce costs and improve safety
§Deliver new vehicle innovation to maintain / extend market differentiation- Particularly software / electronics
§Reduce quality and reliability issues- >50% of life cycle warranty costs
§Ensure security of electronic systems and comply with safety critical engineering process regulations
Business Challenges
§Reduce life cycle warranty costs and improve product differentiation- Reduce dealership visits via OTA software
updates- Improving analytics for early problem detection
§Vehicle-to-vehicle awareness & safety§ Increased driving automated & assistance§Risk-based insurance
Opportunities for Innovation
© Copyright International Business Machines Corporation 2014, All rights reserved 29
Integration to othersmarter systems
Remote diagnosticsand software updates
Vehicle as application platformgenerating new innovation
§Deliver new vehicle innovation to maintain / extend market differentiation- Particularly software / electronics
§Reduce quality and reliability issues- >50% of life cycle warranty costs
§Ensure security of electronic systems and comply with safety critical engineering process regulations
§Reduce life cycle warranty costs and improve product differentiation- Reduce dealership visits via OTA software
updates- Improving analytics for early problem detection
§Vehicle-to-vehicle awareness & safety§ Increased driving automated & assistance§Risk-based insurance
Visualisation of different groups of consumer based on their seasonal usage
© Copyright International Business Machines Corporation 2014, All rights reserved 30
Identifying customers for targeting of energy saving products and advice
© Copyright International Business Machines Corporation 2014, All rights reserved 31
Identifying customers to target efforts to protect potentially vulnerable users
© Copyright International Business Machines Corporation 2014, All rights reserved 32
Wrap-up
Step 1Data Understanding
Step 2Hypothesis Testing
Step 3Reporting
© Copyright International Business Machines Corporation 2014, All rights reserved 33
• Load the data • Explore and sense
check versus expectations
Step 1Data Understanding
• Define hypotheses using expert knowledge and exploratory learning
• Build and iterate models to test hypotheses
• Synthesize analysis results
• Demonstrate possible benefits
• Explain insights• Plan next steps
Step 2Hypothesis Testing
Step 3Reporting
© Copyright International Business Machines Corporation 2014, All rights reserved 34
Questions?Questions?
© Copyright International Business Machines Corporation 2014, All rights reserved 35
Recommended