EVERYTHING THAT MATTERS IN ADVANCED ANALYTICS
Marcia Kaufman, Principal Analyst, Hurwitz & Associates Dan Kirsch, Senior Analyst, Hurwitz & Associates
Steve Stover, Sr. Director, Product Management, Predixion Software
About Us
Hurwitz & Associates -- a strategy consulting, market research and analyst firm focusing on how technology solutions solve real world customer problems.
Marcia Kaufman, COO & Principal Analyst
Coauthor of 6 books including “Big Data for Dummies”
and “Cognitive Computing and Big Data Analytics”
Dan Kirsch, Principal Analyst
Coauthor of “Hybrid Cloud for Dummies” and industry analyst
Agenda
Preparing for Business Change
Hurwitz & Associates 2014 Victory Index for Advanced Analytics
Research Findings & Top Trends in Advanced Analytics
Predixion Victory Index Results
Customer Use Case
Are you Prepared for Business Change?
Finding Opportunity in Sensor and Device Data
Can you combine sensor data with internal and third party data to predict machine failures and improve performance?
Are you reducing fraud by looking for patterns between point of sale (POS) systems, enterprise data sources and demographic data?
Can you improve logistics by analyzing sensor data on shipping equipment in combination with weather, traffic and enterprise data sources?
Hurwitz Victory Index: Advanced Analytics
Assessed advanced analytics offerings of 10 vendors
Surveyed over 450 end users on business and technical value
Performed in-depth customer interviews
Researched customer use cases and best practices
Identified key market trends
Rated advanced analytics
vendors: Vision, Viability,
Validity, & Value
Defining Advanced Analytics
Statistical or data mining solution
Algorithms and techniques used on structured or unstructured data to predict outcomes
Most commonly used techniques
Decision trees
Linear Regression
Time series models
Cluster analysis
Logistic regression
Neural networks
New generation of analytics solutions can hide complexity and add automation
Key Findings: End User Survey
Users finding and
analyzing patterns in unstructured big data sources (i.e. machine sensor, social media)
Users highly satisfied with vendor
Challenges: lack of analytical skills,
integration with other
software, support for in-
database capabilities
Top Trends: Advanced Analytics Victory Index
Analytics platforms becoming more accessible to business users
Visualizations have become a key way to explore data and interpret analytics results
Data sources larger and more diverse increasing
demand for more computational power and in-database capabilities
Increasing demand for pre-packaged analytics solutions
Real-time analysis of sensor and machine data used to anticipate problems and take corrective action
Defining the Internet of Things (IoT)
Network of physical objects
Instrumented with sensors and/or software
When connected, these “things” achieve greater value by communicating with each other or with other data sources
Challenges of IoT data from devices and sensors:
Complex in terms of type (lack of standards)
Expanding exponentially
Flowing at a high rate of speed
Using Analytics to Leverage IoT Data
Monitor temperature, pressure, moisture
Develop predictive models
Detect and analyze hidden patterns and anomalies in sensor and machine data
Common IoT use cases
Predictive maintenance
Customized healthcare recommendations
Streamlined logistics
More responsive IT security
Efficient delivery of public utilities
Go to Market Strength – Predixion has made rapid progress
since its founding. The company built upon its initial success with healthcare to support other industries and use cases including fraud detection, preventative maintenance, and marketing
optimization amongst others.
Customer Experience Strength – Customers like Predixion’s fast time to benefit as well as the integration with R. Customers felt that Predixion’s platform was much more approachable than traditional analytics offerings.
Predixion – A Challenger in the Hurwitz Victory Index
Customer Example: Predictive Maintenance
Predictive Maintenance with the Internet of Things– Intervening to Prevent Oil Rig Failure
Challenge:
A large energy company depends on expensive oil rig equipment to produce revenue
Equipment failure means that operations must stop
Unnecessary maintenance also takes equipment out of use
Solution:
The company used Predixion to build predictive models using sensor data and oil rig maintenance logs – solution created in 3 weeks
Sensors on oil rig components monitor pressure, temperature, vibration
Thousands of scoring predictions are made per second on oil rig components
Field maintenance and operations teams leverage analytics in real time
Key Benefit:
Identify an oil rig failure 14 days before it was set to occur and take
What Users Said About Predixion
“Predixion has been able to differentiate itself by making an accessible tool with a fast learning
curve and intuitive data visualization approach.”
Predixion Partner in the Life Sciences Space
“The Predixion relationship is much more like a partnership than a traditional vendor
relationship.”
Predixion Healthcare Customer
Over 20 years of experience in Big Data and
Analytics, Cloud, and Systems Management markets at market leading companies like Teradata, Red Hat, and Dell.
Led the launch and growth of successful product lines for Cloudera, Oracle, SAP, OpenStack,
Microsoft and VMware.
Steve Stover, Sr. Director of Product Management
• Founded in 2009; HQ in California
• Strategic investors include Software AG, Accenture, GE and EMC
• Expert services team with PhDs in Data Science, Statistics and related fields
PARTNERS CUSTOMERS
About Predixion Software
Predixion named “key player” in Healthcare Analytics Market Forecast (2014)
Predixion included in Big Data 50 – the hottest Big Data startups of 2014
Predixion recognized as a Challenger in Advanced Analytics Market (2014) Predixion debuts on Gartner Magic Quadrant
for Advanced Analytics (2015)
Predixion debuts on Forrester WAVE forBig Data Predictive Analytics (2015)
Industry Recognition
Sources: 1Goldman Sachs, 2Cisco , 3IDC
Devices connected by 20201
Generated by the IoT 2013-20203 CAGR for M2M market1
12B-50B
Devices connected by 20202
44EB-4400EB
Generated by the IoT 2013-20203
40X 60X
Processing Cost1 Bandwidth Cost1
4,400 exabytes = 4,400 billion terabytes
Driving Forces: The IoT Data Deluge
• Optimize key processes:
⁻ Supply chains
⁻ Field operations
⁻ Utilization or consumption
⁻ Customer experience
⁻ Treatment plans
• Preserve the value of equipment by avoiding costly failures
• Automate actions
• Create new opportunities from greater insights into your IoT data
Predict Act
Why Predictive Analytics for IoT data?
Segmentation
Offer Recommendation
Predictive Lead Scoring
Customer Churn
Retailer Fraud
Loyalty Fraud
Readmissions
Length of Stay
Population Health
Claims Fraud
Medication Adherence
Disease Progression
Remote Patient Monitoring
Predictive Maintenance
Field Service Optimization
Inventory Optimization
Predictive Maintenance
Driver Attrition
Driver Incident Risk
ENERGY, UTILITIES & MANUFACTURING HEALTHCARE & LIFE SCIENCES
RETAIL & MARKETING TRANSPORTATION
Predictive Analytics Use Cases
Challenges with Predictive Analytics for the IoT
• Traditional analytics tools cannot handle the volume or speed of IoT data at the edge
• Many devices and machines are in remote areas with limited or periodic connectivity
• Aggregating or filtering the data and forwarding it creates data blind spots
• Critical use cases require real time actions performed at the edge
Taking Action in the IoT – Who, What, and When
Seconds Hours Days Weeks
Executives
Management
Staff
Systems
O P E R A T I O N A L
• Expand capacity
• Alter product line
• Introduce new service
• Schedule maintenance
• Change policies
• Order inventory
• Implement driver improvement
• Repair now
• Schedule patient visit
• Extend production
T A C T I C A L
S T R A T E G I C
• Orderly shutdown
• Next best offer
Predixion provides real-time, predictive analytics at the decision point to improve outcomes.
ACT
API & SDK SOLUTION ACCELERATOR
PREDICT ANYWHERE ANALYZE & MODEL
DATA FROM ANYWHERE
What We Do
CLOUD
SENSORS SMART DEVICES ON DEVICE, IN REAL TIME
• Alerts on both connected & disconnected devices
• Act before costly failures occur
• Automate corrective actions and efficiency optimizations
AT AGGREGATION POINT
• Ideal when on-device deployment is less feasible
• Avoid time lag, data 'blind spots'
• Quickly and efficiently schedule local services
AT CENTRALIZATION POINT
• Enables a broad across the business
• Uncover new business opportunities
• Create new processes and policies
• view to cost optimize
We are the only predictive analytics solution that can run on the device, on the gateway and in the cloud.
Predictive Analytics for the IoT: Where and When You Need it
MODEL CREATIONDEPLOYMENT
DATA
REAL-TIME OR BATCH DATA
• Speed to value
• Portable analytic workflow
• Easy to embed
• Easy to update
RUNTIME
RESULTS
JAVA Combine
Shape Build
Compare Visualize
Collaborate
THE LAST MILE OF ANALYTICS
Patent-Pending MLSM Provides Value in Model Creation & Deployment
Embed Easily into Existing Applications or Create your Own
Portable Web Applications Interactive Mobile Apps
Embedded CRM
Dashboard Integration
Solution Accelerator Embedded IoT
PREDIXION SOLUTION
• Reduce downtime
• Optimize inventory
• Optimize deployment of your technicians
THE CHALLENGE
DESIRED OUTCOMES
• Train the predictive model with sensor data
• MLSM pushed in real time in CEP engine
• 1000s of predictions scored per second in memory with near zero latency
• Predicts individual part failure on individual trucks in advance
• Pushes alerts into Accenture end- to-end fleet management solution
• Quick and easy to embed in a variety of environments
• Can Predixon predict part failures on a large fleet and embed into an end to end fleet management
solution?
Capabilities Proof Point: Predictive Maintenance on Large Fleets
PREDIXION SOLUTION
• Reduce costly failures
• Optimize part inventory
• Optimize field service dispatch
THE CHALLENGE
DESIRED OUTCOMES
• Train predictive models using sensor data
• MLSM Package pushed in real time in CEP engine
• 1000s of predictions scored per second in memory with near zero latency
• Predicts part failure in advance
• Pushes alerts to user friendly UI for action
• Quick and easy to deploy web- based app with our Solution Accelerator
• Can Predixion detect an oil rig failure up to 14 days in advance by analyzing the streaming data from 100s of sensors in the field?
Capabilities Proof Point: Predictive Maintenance on Oil Rigs
• Rapidly create value from your IoT data
• Only predictive analytics vendor that can run on the device, on the gateways or in the cloud
• Expert Data Scientists to help you gain insight and leverage from your IoT data
• Proven ecosystem of technology and delivery partners
Why Predixion Software?
THANK YOU
www.PredixionSoftware.com