Becoming data-driven requires analytics to be embedded throughout the organization in different functional areas and different operational processes. But how do you provide more and more people with the ability to run any analytics on any data anywhere– without breaking the bank? In this session, you’ll see real-world examples of Dell customers who have successfully embedded analytics across processes and operations to drive innovation.We will also demonstrate how embedding analytics enables faster innovation and improves collaboration between data scientists, business analysts, and business stakeholders, leading to a competitive advantage.
5. A complex landscape
Data
• Financial
• Behavioral
• Demographic
• Social
• Text
Systems
• Billing
• MES
• CRM
• HR
• Claims
Regulations
• Internal
• FDA, SEC, IRS
• CPSC
• Safety, State
People
• Managers
• Engineers
• Bbusiness
• IT
Considerations
How can I
use text?
How do I
get started?
Where do I find the data?
Processes
• Claims processing
• Loan approvals
• Customer service
• Purchasing
• Warranty
• Manufacturing
Why is this
so slow?
Has someone
done this before?
Are we
compliant?
Can we do
this more
efficiently?
Why is this hard to use?
Technology
• Cloud
• On-Prem
• Unstructured
• Streaming
6. You can break down the barriers to better analytics.
Develop an analytics
culture
• Embed analytics into LoB
applications
• Improve collaboration
• Identify skill gaps and invest
in training/mentoring
Automate, self-
service and reuse
• Automate tasks
• Add self-service capability
• Use recipes and reusable
templates for best practices
Reduce
dependency on
legacy systems
• Understand hidden
costs
• Use modular,
embeddable
solutions
• Find simple, easy-
to-use solutions
Make data
the
lifeblood
of your
business
Unlock the
power of
all data
7. The changing face of analytics
People
Process
Technology
Embed
analytics
everywhere.
Innovate
faster.
Empower
more
people.
8. Embed analytics everywhere
Data
Demographic
Social
Behavioral
Financial
Predictivemodeldeployed
Predictive Model
PredictiveScoreGenerated
Acts on
“unseen”
data
Buy or click
Churn or defect
Default or
prepay
Contract
disease or
infection, get
readmitted
Commit fraud ,
crime or
terrorism
Likelihood to…
Breakdown or
become
defective
The output
is a score
Machine learning
and data mining
Crunches
data to
build a
predictive
model
Analyticsembeddedintobusiness
Datablending
Dashboards
Mobile
Web
Process
improvement
9. Native Distributed Analytics v1.0 – On premises
Dell Statistica
Statistica Big Data Analytics
Neural Net
SQL Server
Hadoop
Export model as:
1. Java
2. PMML
3. C
4. C++
5. SQL
10. Native Distributed Analytics v2.0 – Scoring
Dell Boomi
Date/Time
Trans type
Velocity
Trigger
Private cloud
JVM
Amazon
Web Services
JVM
Salesforce
JVM
JVM
Dell Statistica
Statistica Big Data Analytics
Neural Net
SQL Server
Hadoop
Export model as:
1. Java
2. PMML
3. C
4. C++
5. SQL
11. Native Distributed Analytics v3.0 – Model build
Dell Boomi
Date/Time
Trans type
Velocity
Trigger
Private Cloud
JVM
Amazon
Web Services
JVM
Salesforce
JVM
JVM
Model build
Model build
Model build
Model build
Dell Statistica
Statistica Big Data Analytics
Neural Net
SQL Server
Hadoop
Export model as:
1. Java
2. PMML
3. C
4. C++
5. SQL
12. Enable across analytics marketplaces and internally with reusable templates
Collective intelligence – analytic workflows
• Internal – how to enable my
workforce?
• Recipes
• Reusable templates
• Embed analytics into
LOB applications
• Shortage of skilled expertise
• Use the global
community for analytic
modules
13. Dell improves attach rates.
Challenge
• Offers were inconsistent.
• There were too many systems and steps required
to pull a quote.
• Recommendations were made too late, after
point of sale.
• Offers were inconsistent with different offers
depending on the sales platform.
Results
• Targeted offer matches customer’s history,
industry and other parameters.
• Automated targeted quote is generated in just
two steps.
• Point-of-sale recommendations made in real-
time.
• Analytics are embedded into sales tools.
14. Large insurance company
reduces fraud.
Challenge
• Find automated way to refer suspicious
claims to special investigations unit.
• Move beyond business rules to incorporate
predictive analytics into processes.
• Incorporate both structured and text data.
Results
• The fraud modeling project paid for itself while
in development. After 2 years the projects ROI
is estimated between 350-500%.
• Claims referred by the modeling project have
an economic impact that averages about 35+%
larger than claim handler referral.
• At current rates, the automated fraud referral
project has increased SIU cost avoidance by
35-50%.
15. ER predicts heart attacks with
near 100 percent accuracy.
Challenge
• ER patients with “potential heart attacks” have
to undergo post ACS testing.
• Overcrowding, clinical risk and duplication of
effort costs billions annually
• Up to 85% of patients are ultimately
diagnosed with other conditions.
Results
• Near 100 percent accuracy compared to
traditional tests
• “A substantial subset of patients [could be]
safely discharged without performing MPS and
subsequent outpatient follow-up”
• Predictive analytics improves patient care
(reduced time and stress) and cuts medical
costs with fewer overnight stays, fewer tests,
and reduced physician time)
17. 3 things to do next
Get inspired. Talk to other Dell customers, visit the Big
Data and Analytics booth to see the demos, and think
about relevant use cases for your organization.
Get going. Assess your analytics maturity to determine
the steps needed to begin embedding analytics
everywhere.
Get prepared. Collaborate with colleagues and analysts
in IT and the business to assess how you can begin to
embed analytics into processes.