View
425
Download
0
Embed Size (px)
Citation preview
© 2016 Information Services Group, Inc. All Rights Reserved.
© 2016 Information Services Group, Inc. All Rights Reserved.
Julien Escribe, Partner, ISG
Reaping the RewardsDigital Business Uses of Data Science
© 2016 Information Services Group, Inc. All Rights Reserved.
Abstract
Data science in all its forms – data learning and mining, predictive analytics, guidance for prescriptive decision-making, machine intelligence, and autonomous systems – is being trialed, tested and deployed for both machine- and people- related business applications, including the consumer and industrial IoT.
This discussion focuses on strategic, tactical, and operational business applications of data science that change businesses and markets.
© 2016 Information Services Group, Inc. All Rights Reserved.
Digital Business is Powered by Data ScienceThe Concept: Data science accelerates business results by using “intelligent digital machines and processes" to further automate market/customer and supply chain interactions, while augmenting human labor & decision-making.
Key Enablers today: Data unique to the enterprise, data external to the enterprise Data mining/machine learning development platforms Intelligent machine processing (IMP) Natural Language Processing (NLP) Robotic process automation / autonomic process automation Adaptable/Reusable IMP software with Spark/Python/scripting/PMML/PFA formats Humans and business processes
© 2016 Information Services Group, Inc. All Rights Reserved.
Agenda
1. Definitions
2. Market Trends
3. Use Cases
4. Best Practices
© 2016 Information Services Group, Inc. All Rights Reserved.
What is Data Science?
Dev-Deployment Platforms: today
Dev-Learning Platforms: today
Data Mining &Machine Learning
IMP + NLP+ RPA + AP
Forecast: years from now
Machine Intelligence
& Reasoning (MIR)
Natural Language Processing: e.g., the human IMP interface, including speech, vision, narrative, sentiment analytics, facial recognition, text, OCR, etc.
Machine Learning: e.g., machines train on data and are used to construct algorithms for learning and making predictions on data
Data Mining: e.g., people (data scientists) discover patterns in data using supervised, unsupervised, reinforcement, deep, neural networks and then reuse/transform for further use Intelligent Machine Processing: e.g.,
machines that have been trained and programmed on data by DM and ML platforms for production use. May contain NLP, RPA and AP. Robotic Process Automation, e.g.,
the production uses of IMP, NLP, and/or RPA To augment/replace a machine/business process
Autonomic Processing: the production uses of adaptable computing – modelled on the human immune system
Cognitive~
Machine intelligence and reasoning: e.g., thinking, reasoning, adaptable, and self-learning machines modeled on the human brain/processing
© 2016 Information Services Group, Inc. All Rights Reserved.
What are the Business Rewards?
Insights and foresight will shape outcomes and industries
Within operating functions, across operations, governance functions, tactical and strategic functions of the organization
Accelerated business performance & industry
shaping possibilities Smart “self service” powered by
machine automation and data Faster market response Improved supply chain efficiency Discovery and ability to take
action on unseen profit pools hidden in silos of data
Increased operating efficiencies
Next generation digital business capabilities
Discovery and leverage of value from data-driven digital business processes
Digital business processes driven by data from within and outside the company
New possibilities for the organization
Man + Machine combination = next-level digital business capabilities
© 2016 Information Services Group, Inc. All Rights Reserved.
How Real Is It Today?Survey Data: > 1/3 of Firms Have the Core Tech in ActionSurvey Question: Rank the following emerging technologies on a scale from “Not Deployed” to “In Production” as they are in use today.
Robotics
Robotic Process Automation (RPA)
Machine learning / AI / Cognitive
Voice Recognition / Natural Language Processing
Data Science / Big Data
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
23%
22%
13%
13%
6%
18%
22%
21%
18%
18%
24%
24%
29%
32%
23%
23%
20%
23%
21%
29%
11%
12%
14%
16%
24%
Not Deployed, No Plans to Use Researching, But No Current TrialsSource: ISG Insights 2016 global IT leader survey; n = 352
The Deployment Reality
© 2016 Information Services Group, Inc. All Rights Reserved.
Vertical industryplatforms
Digital labor/operationsplatforms
Evolution of Data Science in the Enterprise
2000 2005 2010 2015 2020 2025 2030 2035 2040
Businessoptimizationplatforms
Development tools & platforms
Time
Mar
ket i
mpa
ct
© 2016 Information Services Group, Inc. All Rights Reserved.
Digital Labor/Operations Platforms
Machines augment/automate productivity and efficiency gains for operations.
IT OperationsMarketing
Customer service Regulatory...Sales Logistics
Supplychain
Finance/ HR
© 2016 Information Services Group, Inc. All Rights Reserved.
For IT operations, emerging solutions and service provider offerings are being deployed to improve operational awareness, provide straight-through processing for incident resolution, and shift-left for organizational effort dedicated to IT operations.
Job Monitoring
DB & MW Monitoring
Hypervisor Monitoring
Desktop Manageme
nt
Next-Gen IT OperationsTicketing & Service Desk
CMDB
Process Automation
Network Manageme
nt
Advanced Event
Correlation
User Experience Monitoring
CI Visualization (Faster RCA)
Add’l Process Automation
Dashboard & Reporting
Decision Engine
(ML Self-Heal)
Security Monitoring
Server Automation
Server & App
Monitoring
Backup & Storage Mgmt
RTPaaS Monitoring
Project Requirements & Outcomes: Mashups of various solutions/software >$$$ in labor/non-labor deployment fees Scripting team writes knowledge items Six to twelve month deployments are
typical Business case savings of >20% of TCO
Example of Digital Labor/Operations in IT
© 2016 Information Services Group, Inc. All Rights Reserved.
Intelligent machines discover profit opportunities from operational silos of data.
Optimizes decisions to maximize pricing/market mix/customer yields/inventory/distribution/supply chain.
Resources and costs are optimized.
Market and profit opportunities are accelerated.
Business Optimization Platforms
Sales &marketing
ProductionSupply chain
Logistics, distributionand customer service
Business OptimizationPlatform results
© 2016 Information Services Group, Inc. All Rights Reserved.
Vertical Industry Platforms
IMPs augment human processes and decision-making to improve agility, efficiency, quality, customer retention, field trials, patient compliance, healthcare diagnostics… for industry specific uses.
OperationsLegal...
Industrials,ProductionTransportGovernment
Retail &Consumer
BFSIInsuranceCPGMedia
© 2016 Information Services Group, Inc. All Rights Reserved.
Example of Vertical Industry Platform: Fraud loss Detection/Prevention in Insurance Industry
For agent/broker operations, protecting the company’s loss ratios are more important than ever.Emerging solutions and service providers help companies protect their customers and investments against fraud by utilizing cognitive technologies.
Claim Filed Storage Analysis
Stop Payment, Report to Authoritie
s
Fraud
Risk Portal
© 2016 Information Services Group, Inc. All Rights Reserved.
Sample IT Providers
Open Source DevOp platforms Apache Spark, Apache Mahout, H2O, Kaggle, KNIME, LIBLINEA, MLLib, Octave, OpenNM,Orange, PredictionIO, R, RapidMiner, Tangara, Scikit-Learn, SciPy, Sense and Weka among others
Commercial DevOp platformsActian, Amazon, Angoss, BigML, CognitiveScale, Dato, Dell, DigitalGenius, Fuzzy Logix, Google, HPE, IBM, Lavastorm, Microsoft, Oracle, Pivotal, Predixion, RapidMiner, SAP, SAS, Skytree and Tibco among others
Digital labor/Operations platformsActionIQ, ADP, Altryx, Aragon, Ariba, Aviso, Bottlenose, Clarabridge, Blueriver, Brightfunnel, ClearStory Data, Crimson Hexagon, Dataminr, DecisionIQ, Deep Genomics, DigitalGenius, Entelo, Everlaw, Gainsight, Framed, FICO, Fuzzy Logix, Honeycomb, IBM, iCIMS, Infer, Infor, InsightSquared, Inventure, IPSoft, JDA, Kana, Lavastorm, Liftigniter, Mattermark, Microsoft, Miradoor, NetMap Analytics, Oracle, Owlin, Predicsis, PTC, Quantfind, Qubit, Radius, Ravel, Salesforce, Salespredict, SAP, SAS, Sentient, Sense, SugarCRM, Taleris, Textio, Workday, Zaius and Zoho among others
Business Optimization platformsAdvanced Software Applications, Applied Predictive Technologies, ClearStory Data, FICO, IBM, R4 Technology, SAS and SmartOrg among others
Vertical industry platformsIBM, Quid, RiskIQ, SmartOrg, Strategic Analytics and TrendSpottr among others
© 2016 Information Services Group, Inc. All Rights Reserved.
Overcoming Some of the Challenges
Some of the challenges Overcoming the challenges Let’s just buy an off-the-shelf software
packageData science is the new data-driven DevOps. It is not an off-the-shelf software package - yet.
Data science finds the sacred cows of the business
Use its flashlights/floodlights to improve results. It is about people and process, more than technology
Process reengineering is more difficult than the data science
Embrace it: data-science driven business is digital business process reengineering
Focus on what the data says is possible and not the business
Digital business is informed by data – but it needs to be channeled by the business
© 2016 Information Services Group, Inc. All Rights Reserved.
Best Practices
1. Define the business opportunity: do not boil the ocean
2. BPR: embrace data enabled business reengineering
3. KYD: know your data sources and limitations
4. Experiment: it’s a combination of art and science
5. Use a fan-out organizational structure: centralize scarce data scientist/programmer/ data engineering resources, fan out using business analysts as conductors-liaisons, embed with SMEs everywhere
www.isg-one.comimagine your future™
let’s connect…