from data to action: driving innovation through predictive analytics (166985443)

7/29/2019 From Data to Action: Driving Innovation through Predictive Analytics (166985443)

http://slidepdf.com/reader/full/from-data-to-action-driving-innovation-through-predictive-analytics-166985443 1/65

Analytics in Higher Education

Vince Kellen, Ph.D.Senior Vice Provost

Academic Planning, Analytics and Technologies

University of Kentucky

[email protected] September, 2013

This is a living document subject to substantial revision.

mailto:[email protected]

mailto:[email protected]



Analytic categories

Helping the student

• Personalization and interaction

• Learning analytics

• Recruiting and retention modeling

• Micro-surveys

Helping the university

• Building utilization and course/event planning

• Monitoring and predicting enrollment in classes

• University research and instructional output

• Tuition revenue projections

2



Models, models, models

Student enrollment

Student recruiting

Student statistics per term

Student demographics and

demographic details

Student accounting and student

financial aid

Student retention cohorts

Student micro-surveys

3

Student majors/minors

Student advising

Student K-Score

Transfer students

Faculty statistics per term

Building utilization

Human resources statistics

Key business process workflow



Poll

Rate your data integration maturity

1. We integrate data in nightly batches from some data sources

2. We integrate data in nightly batches from all significant data

sources

3. We can integrate data real time from some data sources

4. We can integrate data real time from all significant sources

4



Volume operations versus complex systems

Excluding the late 20th century, universities have been largely complex systems, delivering niche and

customizable interactions F2F settings. Large lectures were added to increase output while reducing costs

Analytics, technology and personalized e-Learning approaches can begin to handle both high-volume and

specialty lines of business

5

Geoffrey Moore (2005). Dealing With Darwin



6



Academic Health Notifications:

View in student mobile app



Tallycats

Academic HealthBar (Alerts)

Bronze

• Academic Health Bar only decrements when a specific adverse event is detected.

• Students start in a “Healthy” state, and accumulate decrements for each adverse

event.

• A balanced notification system should be designed such that many students

never receive adverse notifications and thus never even view the Health Bar

screen of the mobile app. We do not want students feeling nagged by the system.

Healthy At RiskUnhealthy

Academic Health Notifications

All students start here



Academic Health

Bar (Alerts)

• The key to the system as a tool for positive behavior is that each notification includes

relevant information and/or necessary actions.

• For low severity events, the notification may merely suggest resources that may help,

in which case the student can manually dismiss the notification.

• More severe notification types will have clear and specific actions the student can taketo clear the notification.

Healthy At RiskUnhealthy

The Blackboard system shows you have submitted only 1 of the last 5

assignments for MA109. The class average is 4.6 submitted assignments.

To clear this alert:

• Either

-Meet with your advisor

-Or meet with your professor

-Or attend an appointment at the Study or Mathskellar • And submit an assignment



15



Advising Hub: Home Screen



Poll

What analytic technologies do you employ (select all that apply)

1. We use distributed analytic approaches (e.g., Hadoop)

2. We use relational database analysis tools (e.g., from vendors like

Microsoft, Oracle, IBM, etc.)

3. We use high-speed in-memory methods (e.g., SAP HANA)

4. We use social media analysis tools

17



WAKE UP! GET TO CLASS!

Who sets alarms for themselves?

Why not automatically set alarms for

students around their schedule?

Why not have automated wake-up

calls?

Why not suggest wake up times

based on class attendance?

Why not consider manipulation of

reminders as a form of engagement?

Can we ascertain student prospective

memory capability and personalize

based on it?

18



What would Abraham Lincoln think of a MOOC?

Abraham Lincoln

• Autodidactic

• Books, books, books

• Became a skilled military strategist

• Penchant for poetry, Shakespeare,

politics and history

My nephew

• Not an autodidact

• Good worker, smart kid, but…

• It takes a village

• After a few low-security colleges

and much money borrowed

• He has found an intellectual home

19

http://blog.cutter.com/2012/09/08/educating-abraham-lincoln/

http://blog.cutter.com/2012/09/08/educating-abraham-lincoln/



Assumptions

The rewards of investments in teaching technology go

disproportionately to the most capable individuals

• The most capable learners tend to receive more benefit from

technological enhancements than less capable learners

Choosing college is a complex decision

• Cultural, social and economic factors affect the decision

• Wealth stratification plays a role (and it has for 2,300 years)

• It isn‟t an entirely rational decision

Information technology (IT) does matter • IT can improve productivity

• IT can enhance short-term and mid-term competitiveness

• But, advantage from IT does not follow its procurement, but from its

marrying with business processes and organizational capital

20



Small class, large class, books & MOOCS

Small class Large class Book MOOC

Student-teacher

interactionHigh Low Low Low

Student-student

interactionHigh Low Low Low-mid

Quality of

instructor Low-high Low-high Moderate-high Moderate-high

Convenience Low Low High High

Overall

experience ?? ?? ?? ??

21

http://www.nytimes.com/2013/04/21/opinion/sunday/grading-the-mooc-university.html

Conclusions about the quality of the experience need to take into account what the

learner is capable of and what they need



Déjà vu?

22

https://reader009.{domain}/reader009/html5/0421/5ada9864b514e/5ada98707abc6.jpg

MOOCs

Large

lectures

PHI 698

???



How technology can affect cost and quality

23

High

effectiveness

Low

effectiveness

Low volume High volume

Small F2F class

Broadcast class

Current MOOC approach

MOOC + PT

MOOC + PT + F2F

F2F = Face-to-face

PT = Personalization technology, adaptive

learning technology



Personalization technology

• Real-time personalized interactions• Target on-demand peer tutoring based on student‟s profile

• Deliver micro-surveys and assessments to capture additional information

needed to improve personalization

• Give students academic health indicators that tell students where they can

improve in study, engagement, support, etc.

• Let students opt their parents in to this information so the family can support

the student

• Tailor and target reminder services, avoid over messaging

• Allow for open adaptive learning• How content gets matched to students is psychologically complex

• Several theories of how humans learn give many insights• Students differ in the following abilities and attributes: visual-object, visual-

spatial, reasoning, cognitive reflection, need for sensation, need for cognition,

various verbal abilities, confidence, persistence, prospective memory, etc.

• We need an open architecture to promote rapid experimentation, testing and

sharing of what works and what doesn‟t




Potential services

Reminder services

• Upcoming classes, assignments, tests, events

Predictive support

• Micro-segment prediction of academic difficulty, involvement, integration

• Non-cognitive factors, social network analysis in models

• Behavioral cues around registration, financial aid applications

• Digital footprint analysis (use of degree progress tools, academic „category‟

involvement, walking across campus)

• Recommend friends, study groups, student groups, foster peer-interactions

Real-time analytic integration with LMS content consumption

• Detect lack of comprehension, difficulty, frustration

• Recommend peer tutoring, additional materials, tutoring services

• Call students in the middle of?

Parent portal

• With the student’s permission, let the parent see key learning performance indicators,

alerts, etc.

25



Poll

Rate your organizational readiness for analytics

1. Various units work independently on different types of analysis

2. We have occasional coordination across multiple units that

perform analysis

3. Multiple units across campus share a common data warehouse for

single-source-of-truth data

4. Multiple units across campus readily share analysis techniques,

artifacts and tools with each other

26



Taxonomy? Automatic metadata? Automatic

atomic metadata?

Let learners navigate an

audio/visual stream

Let the system learn what are top

terms. Let the system map terms

to concepts. Let instructional

designers lightly „bump‟ the

taxonomy, post production

Record student engagement with

specific terms / concepts

Deliver personalized messages to

students 27



28



Class slides take a

central position.

The audio/video and

slide content is

synchronized.



The note pad allows

recording and

sharing of notes.



Personalized

recommendations are a

guide thru the material.

Feedback on engagement

and mastery assist in

gauging understanding.

Resources and tutors arealso provided if a little

assistance is needed.



A lecture concept map helps to put the

lecture in a visual context.

The map is generated from analysis of

the text and ‘bumped’ into shape by a

course designer or instructor.

Concepts can be rated to collect

perception of usefulness and improve

future versions.

Jump to the media segment by clicking

on the tag.

Test knowledge with small quizzes – ace

the quizzes and you’re in good shape!



A one-stop-shop for searching.

Keywords from the video,

slides, trends, notes and

conversations will appear.



Key questions

• Can the audio and slides be reliably transcribed into ‘useful’ text?

• Can a concept map be derived automatically from the text

generated or easily edited by an instructor?

• How easy will it be for designers-instructors to create a quiz and

place it in the right location in the video?

• Can we personalize the recommendations to reflect prior

knowledge, student ability and individual differences in

information processing?

• Can the interface support real-time integration with analytic back-

ends (e.g., HANA)?

This is just one conceptualization.

What other interface designs exist today? How effective are they?



Poll

Have you integrated your mobile and analytic strategies? (check all

that apply)

1. We are well underway in deploying analysis tools to smartphones

and tablets

2. We are gathering mobile interaction data to understand studentand user behavior

3. We are exploring mobile deployment of analytic tools to

smartphones and tablets

4. We not ready to gather mobile interaction data into our analytic

environment



The debate

Constructivists/social learning versus reductionists/cognitive psychology

• „Mind as sacred space‟ versus „brain as decomposed machine‟

• Learning as a social-emotional interaction with humans versus learning as

cogno-neuro-sensory interaction with information

Social constructivist, subjectivist, humanist

• I as teacher am the one who personalizes things• Students construct their knowledge via indeterminate, emergent social

interactions

• I don‟t see how IT has a significant role in this, except to support ease of

student interaction with each other and with content

Cognitive reductionist, positivist, technocrat

• An IT system can on the fly determine which content/people you the learner

may need next, learning follows principles of progression

• Students differ in the cognitive skills in ways IT can address

• IT can be a more complete feedback-driven learning environment for all

38



Questions?

39



Innovation in Information Management:Key Considerations around Data & Predictive Analytics for Higher Education

Jon Phillips

Dell Managing Director - Worldwide Education



41 9/10/2013 Software

Information Management

Database management

Application & dataintegration

Business intelligence andBig Data Analytics

Integrate datadisparate data stores,

cloud and on-prem

Inform decision makers withanalysis based on all types

data sources

Manage data and databasesacross structured and non-structured data sources -

cloud or on-premise



42 9/10/2013 Software

Information Management

Database management

Application & dataintegration

Business intelligence andBig Data Analytics

• Foster IT and LOB collaboration

• Capitalize on existing investments

• Rapid time to value

Datalocationagnostic

Datatype

agnostic

Vendor agnostic

Th V l f B i I t lli M t it



43 Services ConsultingConfidential

Stage 2:

Inception

Stage 3:

Refinement

Stage 4:

Consolidation

Stage 5:

Enhancement

Stage 1:

Baseline

Stage 6:

Innovation

The Value of Business Intelligence Maturity

Operational Data Spread-marts Data MartsData

WarehousesEnterprise Data

WarehouseAnalyticServices

Cost Center InformExecutives

Empower Workers

Monitor Performance

Drive theBusiness

Drive the Market

BI Unaware Shadow ITMinimal

EngagementIT Executive

Sponsor Business Exec

SponsorsHolistic BIStrategy

BI Maturity

Maturity increases value & decreases cost

St f B i I t lli M t it



44 Services ConsultingConfidential

Stage 2:

Inception

Stage 3:

Refinement

Stage 4:

Consolidation

Stage 5:

Enhancement

Stage 1:

Baseline

Stage 6:

Innovation

Stages of Business Intelligence Maturity

One-off Reports Spread SheetsOperationalDashboards

Line of BusinessDashboards

BusinessProcess

Dashboards

Customer Facing

Dashboards

Operational Data Spread-marts Data MartsData

WarehousesEnterprise Data

WarehouseAnalyticServices

Cost Center InformExecutives

Empower Workers

Monitor Performance

Drive theBusiness

Drive the Market

Operational Data Stores On-line Analytical ProcessingEnterprise Data

ModelBig Data

Defined ToolsetMultiple Tools

Meta DataCumbersome Data AcquisitionLife Cycle

ManagementAutomated Data

Quality

Stove Pipe Architectures Robust & Scalable Architecture

Project Based Development 5 Year Roadmap

Steward less Minimal SkillsTactical

StewardshipFederatedProcesses

ReportingFramework

BI StrategicCompetency

BI Unaware Shadow ITMinimal

EngagementIT Executive

Sponsor Business Exec

SponsorsHolistic BIStrategy

BI Centers of Excellence

Enterprise DataGovernance

Analytics Center of Excellence

Ungoverned Minimal ProcessPockets of

Governance

AnalyticalCompetency

PrescriptiveAnalytics

Raw DataData

VisualizationInformationDiscovery

DescriptiveStatistics

New ParadigmHistorical Norm

Moving forward to the New Paradigm

Road blocks and capabilities to build at each level of maturity



45 9/10/2013 Software

Road blocks and capabilities to build at each level of maturity

Spreadsheets + lack of systematic analysis

Budget constraints + poor

data integration + poor dataquality

Slow and lengthy BIprojects + poor datainteractivity

Limited skills and executive

support

Under-developed tools andlimited skills

What’s lacking

Predictive

analytics

Proactive reporting

Information consolidation

Casual data access and analysis

Cognitive

analytics

Pivot tables, operational

reporting

Data modeling, Database anddata warehouse design

Data quality, master data

management

BI & dashboards, KPI metric

design, data governance

Data mining, forecast

modeling, trend metrics

Social listening, sentiment

analysis, pattern recognition

Capabilities tobuild

Biggest inhibitors to realizing value of big data



46 9/10/2013 Software

Biggest inhibitors to realizing value of big data

30% 30%

22%

13%

5%

Leadership andorganization

Analyticscapabilities and

skills

Infrastructure andarchitecture

Investment,budget, ROI

Risk concerns(security, privacy)

4.4 M big data jobscreated globallythrough 2016. Only1/3 will be filled

Unstructured data

hard to analyze

Raw-data-to-insights

latency (time-to

action)

Need to exploit

diverse content

RDBMS/traditional

infrastructure won’t

work

Survey data Gartner – May 2012

Historically data management was based solely on structured data



47 9/10/2013 Software

Historically, data management was based solely on structured data

Trained Staff

Tool

Data Type

Database ETLData

WarehouseData

ProvisionData

Analysis

Structured

The explosion of data continues to burden the data tool chain



48 9/10/2013 Software

Petabyte

Exabyte

Zettabyte

Terabyte

The explosion of data continues to burden the data tool chain

Transactional Data

Traditionally, onlytransactional data was

generated and stored in

databases

• Structured

• Measured growth

Human FilesBut over time, westarted creatingunstructured data

• Likes, tweets,relationships (social)

• Sensor data (machine)

• Exponential growth

Social & Sensorshave added

exponentially

mainframe PC internet mobile machine

• Docs, Images, Video• Multiple formats

• Fast growth

Resulting in a complex environment today



49 9/10/2013 Software

Resulting in a complex environment today Volume/velocity/variety and disparate systems have fractured the tool chain

Structured

Text

Sensor

Social

DB ETL DP DA

DB ETL DP DA

DB ETL DP DA

DB ETL DP DA

Trained Staff

Tool-Chain

Data Type

DW

DW

DW

DW

Structured

DB ETL DP DATrained Staff Tool-Chain

Data Type

DW

Analysis

Analysis

Analysis

Analysis

Analysis

Trained Staff

Tool-Chain

Data Type

Trained Staff

Tool-ChainData Type

Trained Staff

Tool-Chain

Data Type

The needed approach minimizes the complexity and enables analyzing all data



50 9/10/2013 Software

• Innovative

Informationmanagement fosterscollaboration betweenIT and LOB

• Analyze all data toyou can break down

the data silos andadvance your business objectives

• Empower morepeople to discover opportunities byoffering self service

access to data

The needed approach minimizes the complexity and enables analyzing all data

Structured, Text, Sensor andSocial

DBETL

DP DADW

Analysis

One Vendor – Complete Tool Chain - All Data

The Big Data ecosystem



51 9/10/2013 Software

The Big Data ecosystem

M o v

i n g U p t h e S t a c k

Hardware InfrastructureServers, Storage and Networking

Information Management/ Ent. Search Architectures, policies, and process, for data lifecyclemanagement. (controlled, optimized, access, searched)

BI ToolsDashboards, Reporting

ETL and Data IntegrationBulk, Real-time

Advanced AnalyticsPredictive Modeling, Data Visualization tools

Database Management SystemOLTP, Relational, SQL, NoSQL, NewSQL, Columnar, Pig

Data WarehouseOLAP, Hive

Serviceconsiderations

Consulting

Implementation

Support

Maintenance

Training

Big Data Hosted

Services



52 9/10/2013 Software

Dell’s Point of View

on Analytics

Education Analytics for Higher Education



53 Copyright 2013 Dell Inc.Prepared by

5 Confidential

During a five-year period,

taxpayers spent over $9 billion to

support college students whodropped out before their

sophomore year.1

• Recruitment and retention

• Sound fiscal management

• Operational efficiency

• Efficient and reliable complianceand reporting

• Grant success tracking

• Monitoring of PK-20 data (all of education plus the workforce)

• Gainful employmentMeasurement / WorkforceReadiness

53 Confidential Global Marketing

Education Analytics for Higher Education

“Saving 1 percent in Student

Retention can save my

University’s bottom line

$1M a year”

CIO Major University

1. American Institutes for Research (AIR). http://www.air.org/news/index.cfm?fa=viewContent&content_id=988

http://www.air.org/files/AIR_Schneider_Finishing_the_First_Lap_Oct101.pdf

What is your BI and Analytics roadmap?



54 Confidential Services

POC (opti onal) • Specific Use Case

• Produce new businessinsights

• Demonstrate data &technology feasibility

Readiness

Assessments

• Strategy and Roadmaps

• Data readiness

• Technology readiness

• Business cases

21

Our Key Offerings

Technology Capabilities Sharing Key Experiences

Major universities, large banking and financialinstitutions, large healthcare systems, keydata distributor, etc.

need actionable

information

Educators need new

business insights

Decision making

requires holistic view

of enterprise data

Existing DW and

BI investments need

to be leveraged

Analytics Engines: SAS, R, SPSS Analytics Delivery Platforms: MS BI, SAP HANA & BOBJ,

SharePoint

Data Visualization Tool: Tableau, QlickView, MS PowerView

Data Engine: SAP HANA, MS SQL, Oracle

Solut ion

Implementat ion

• Model DLC • Leverage Dell DW and

BI delivery capabilities

• System Integration

3

Advanced Analytics Offerings

Managed Servic es • Continuous model

improvement • Subscription model

• Expanded use cases

4

Ideal solution approach



55 9/10/2013 Software

Ideal solution approach

Analytics

Sophisticated NLP, machine learning, predictive modeling,

sentiment analysis, SNA, and visualization. No

Hadoop/MapReduce programming expertise required

Analyze

unstructured,

structured and

semi-

structured datafrom a single

work-bench

Search

Interactive and intuitive. Search interface allows businessanalyst to explore and exploit all data resources

Visualization

Interactive web-based authoring empowers business users

to perform analysis, visualize results and take decisions

Analytical

Producer

Analytical

Consumer

Capabilities



56 9/10/2013 Software

Capabilities

Natural languageprocessingDerive value from the vast

sources of unstructured text and

scanned content from archives

Entity extraction

Easily extract entities from rawunstructured text.

Machine learning

scalable machine learningprovides the ability to trainalgorithms over time for predictive analysis and pattern

recognition

Social network analysis

Ability to identify unknownrelationships in social networks.

Search

Leveraging industry' powerfulsearch platform with full-textsearch, hit highlighting, facetedsearch and real-time indexingprovides insight into largevolumes of data.

Sentiment analysis

Decipher true meaning behindunstructured text.



57 9/10/2013 Software

A case for advanced

Analytics

Student Retention

Dell Technology Helps College Students



58 Copyright 2013 Dell Inc.Prepared by

Learning Management System Ecosystem

Dell Technology Helps College Students

Transition to a Successful College Experience

Assess

Risk

Inform

Student &

Faculty

Educate& Intervene

Take

Action

Integrate

Data

BI Analytics

Dell Student

Retention Model

& Products



59 Copyright 2013 Dell Inc.

Prepared by

Dell Student Retention Model

Dell Student

RetentionPredictive

Analytics Model

StudentAdvisor

Reports &

Alerts

Reports &

Alerts

Model

Variables

Early Intervention

Risk

Factor

• Ethnic Group

• First Generation Flag

• Gender

• Resident Status

• High School GPA

• ACT/SAT Score

• Current GPA

• Has PELL Grant

• Campus Activities

• Student Data

• Course Data

• School & Faculty Data

S-Score

Advanced BI and Analytic Reporting




Prepared by

Advanced BI and Analytic Reporting

• Exec Leadership

• Administration

• Financial Aid

• Institutional Research

• Departments

• Faculty

• Students

• Financial Aid

• Counselors

Dell Student Retention: Data -> Insight -> Action



Use Case: Automatic rule-based alerts for contact initiation

LMS DB

DSR Rules

DSR DB

Email Alert

1. New academic

activities imported

into DSR

DSR Model

3. Student Risk

Score is published

4. Student‟s risk score

reaches alert threshold

5. Email/Text alert sent to

Advisor and Student

2. Student Risk

Score is updated

DSR BI

6. Advisor

receives

risk report

Calendar 7. Advisor sends out

iCalendar

8. Student accepts

meeting invite




Prepared by

Cultivating Student Success with Insights

Marketing Recruit AdmissionFinancial

Aid

Enrollment Education Retention GraduationAlumni

Outreach

Targeting

Success

Potential

Address

Financial

Risk

Risk

Assessment

Program

Planning

Course

Planning

Select

Effective

Approach

Closely

Monitor

Measure &

Adjust

Following

Career

Success

Exec ManagementInstitutional Research

Student Retention Model Adoption Process




Prepared by

Analytics Development

Lifecycle (ADLC)

Student Retention Model Adoption Process

Dimensional

Student Retention

Model

Student

Demographics

Course

Enrollment

Historical Cohort Data

Student

Grades

Other

Activities

Scores

(Risk & Performance)

Alerts

Actions &

Improvements

Dimensional DataIntegration

Variable

Mapping

ModelValidation

ModelDeployment

ModelMonitoring

Model

Tuning

BI Development Lifecycle

Moving Beyond Prediction- Advanced Analytics in Action



64 Confidential Services

OriginalPrediction

NewTarget

Why How

• Target driven analytics

• Focus on changingthe outcome

Prediction

• Traditional approach

• Focus on prediction

• Limited to known facts

Target

• Advanced Modeling

• Actionable Levers

• Testable & Repeatable

• Alerting business outcome

• Testable & repeatable results

• Operationalization

NewReality

A c t i o n s

Lift of BusinessOutcome

Prediction

New ParadigmCurrent Norm

• Lever -1

• Lever -2• Lever -3

Success

1

2

3

Open Integration with Your Existing Systems



Dell

Student

Retention

Product

p g g yProduct Neutral, Customizable, Complete Stack

*.edu

Data Adaptors

Student Retention ModelOther BI Platform

Dell QSDW

from data to action: driving innovation through predictive analytics (166985443)

Documents