Business Analytics Forum - Sas Institute · “Percieved camper safety can be impacted by the level...

Copyright © 2011, SAS Institute Inc. All rights reserved.

Business Analytics Forum 02NOV2011 Jim Metcalf Sr. Director of Testing Advanced Analytics Division

2


This is what Business

Intelligence looks like…

3


But THIS is what SAS analytics can do for you now …

4


Why “analytics”?

What happened?

How many, how often, where?

Where exactly is the problem?

What actions are needed?

Why is this happening?

What if these trends continue?

What will happen next?

What‟s the best that can happen?

5


Three Use Cases

SAS Enterprise Miner

SAS Text Miner

SAS Forecast Studio

6


Analytics in Action using SAS®

How can college admissions offices identify the best-matched student candidates while speeding response time to the most-

appropriate students….and lower costs at the same time?

7


8


Founded in 1973

1,600 colleges and universities serve as customers

Leader in college and university enrollment management solutions

Student recruitment

Student retention

Financial aid management

9


Critical Business Drivers

Provide intelligence to help colleges:

• Identify “best-matched” prospective students

• Improve academic profile

• Increase enrollment

• Maintain and/or increase academic quality

• Identify students most likely to matriculate

• Increase net tuition

Expedite pipeline timeframe from initial search to decision.

10


Potential student qualification

Saves money by eliminating the need to communicate equally with every prospect/inquiry/applicant

Facilitates better relationship-building by truly personalizing admissions process for those students with the greatest propensity to enroll

Focuses admissions outreach efforts (travel)

Provides a mechanism for enrollment forecasting

11


Traditional means of qualifying students

Written communications (reply cards)

One student at a time

Phone call to student

Track student contacts….manually

12


The Solution - Predictive modeling

Statistical analysis of past enrollment behavior to predict future enrollment behavior

In this case, the likelihood that a student will enroll is determined by the degree to which the student shares the characteristics of the current student body (or a subset of the student body)

13


Better management of the marketing/recruitment process

Segment and customize written communications

Segment and prioritize telephone communications

Segment and prioritize travel

Reallocate limited resources strategically

Increase enrollment by targeting hottest segment

Shape the incoming class

Save thousands of dollars for other enrollment expenses

Predictive modeling

14


Sample model parameter inputs

15


Initial model creation

Client PC

Client mainframe student data

Microsoft Internet

Explorer (XML)

Microsoft IIS

Web Server

SQL Server

Person Warehouse

with Axciom data

Noel-Levitz

Statisticians

Models deployed

To SAS Scoring Server

SAS® Enterprise Miner

16


Real-Time Scoring Process

Client‟s

Recruitment

Application

SOAP Model called with parameters

Microsoft IIS Server

SAS

Web Service

SQL Server

Person Warehouse

With Axciom Data

SOAP Score returned from model

Using this topology, can score 50,000 names in 3 minutes

Models deployed

To SAS Scoring Server

http://images.google.com/imgres?imgurl=www.claybennett.com/images/archivetoons/olympic_judging.jpg&imgrefurl=http://www.claybennett.com/pages/olympic_judging.html&h=300&w=423&prev=/images?q=olympic+judging&svnum=10&hl=en&lr=&ie=UTF-8&oe=UTF-8

17


Client

Noel Levitz clients using .NET to Enterprise SAS Servers

IOM Bridge for COM

z/OS Series

UNIX

DCOM

Windows Server

2008 R2

Microsoft IIS Server

SAS

Web Service

SOAP

http://images.google.com/imgres?imgurl=www.girlzilla.com/images/ipaq.gif&imgrefurl=http://www.girlzilla.com/work_games_demos_rokkett.asp&h=394&w=256&prev=/images?q=pocket+pc&svnum=10&hl=en&lr=&ie=UTF-8&oe=UTF-8

18


ForecastPlustm Return on Investment Reduced cost to implement with SAS was $200K instead of

$600K…a $400K savings

Instant student scoring, instead of 5-day wait.

Improved customer satisfaction

Increased Sales

Significant competitive advantage

With ForecastPlustm their customers have significantly increased

Student retention

Academic profile

Enrollment

Net tuition

19


Raised freshman enrollment by more than 3,000 students, an increase of nearly 40%. University in the West

Enrolled the largest incoming class since the 1960s. Another University in the West

Increased applicant pool by 35% from the previous year. University in the Midwest

Reduced inquiry pool by 27%. University in the South

Enrollment increased by 34%. University in the South

Enrolled the largest incoming class in its history College in the Southeast

ForecastPlustm Return on Investment Customer-driven successes

20


Alberta Parks use of SAS Text Mining Technologies in 2011

Alberta Tourism, Parks and Recreation

21


22


Alberta Parks

Alberta is a world-class tourism destination with active, healthy citizens who participate in sport and recreation and value their parks and natural heritage.

The Parks and Protected Areas Division manages nearly 500 sites covering 27,669 square kilometers in Alberta's network of parks and protected areas.

These preserve important ecological areas and provide places where people can enjoy and learn about Alberta's priceless natural heritage.

23


Alberta Parks Public Engagement

Alberta Parks is committed to engaging Albertans in decisions about their parks. To fulfill that commitment, we have a framework that describes when and how we will consult with Albertans.

There are a number of methods used to collect feedback...

Surveys

Emails

Mail (i.e. letters)

Website comments (e.g. blogs)

Public Consultations (i.e. face-to-face)

Other Social Media Platforms (coming soon)

24


How Alberta Parks views Text Mining

Text Mining:

“Application of

Information Retrieval

and Data Mining

techniques that

accommodate text as

an input variable in

knowledge discovery

or predictive

modeling”

Analyzing open-ended survey responses

Automatic processing of messages, emails, website comments, twitter feeds and letters from the public

Investigation of web site visitation patterns

25


Example: Customer Satisfaction Surveys

Every year Alberta

Parks conducts a

survey of overnight

customers (Campers)

and asks them: What

could we have done

to make your visit

better?

26


27


28


Alberta Parks: Customer Satisfaction Surveys

Their hand written

responses are not ideal.

It comes with the nature

(no pun intended) of their

business….Happy

Campers!

Comments are (currently)

typed in to Excel. In the

future, they will be

transferred to electronic

format using Speech to

Text (dictation) software.

29


30


31


32


33


Alberta Parks: Analysing Text

The Old Way

1. Typing comments

(~3 weeks/year)

2. Every comment

manually read and

manually assigned

special codes (~ 3

weeks/year)

The New Way

1. Dictation software

types comments (~1

week/year)

2. SAS Text Miner

analyses data (~ 1

minute/year)*

*Time savings is

significant.

34


Results from the Old way

Once comments are

assigned codes, simple

frequency counts show

magnitudes of customer

feedback…

…see example of

frequency counts on

the next slide…

35


Results from the Old way

36


Alberta Parks: Results from the New way

SAS Text Miner

discovers

relationships.

This is much more

valuable and positions

us to shift from reporting

the past to anticipating

the future.

An example of results

can be found here:

http://www.albertapar

ks.ca/pubsandmedia/

2010_prov_summary.

pdf

(http://albertaparks.ca/publications.aspx#research)

Example: in 2010 Alberta Parks

discovered a link between a

campers perception of safety and

other services provided. It was

reported that:

“Perceived camper safety can be

impacted by the level of noise,

bathroom or site cleanliness, and

the amount of officer patrols.

Failing in any of these may

contribute to campers feeling

unsafe”.

http://www.albertaparks.ca/pubsandmedia/2010_prov_summary.pdf




http://albertaparks.ca/publications.aspx

37


Alberta Parks From knowledge to actionable intelligence

There is no way to determine

outcomes like this:

“Percieved camper safety can be

impacted by the level of noise,

bathroom or site cleanliness, and

the amount of officer patrols.

Failing in any of these may

contribute to campers feeling

unsafe”.

From output like this:

38


Alberta Parks From knowledge to actionable intelligence

39


Jared Prins BSc. , Program Analyst

Business Integration and Analysis Section Alberta Tourism, Parks & Recreation

2nd Flr. Oxbridge Place, 9820-106 Street

Edmonton, AB

T5K 2J6

Phone: 780.427.6313

Fax: 780.427.5980

www.AlbertaParks.ca

Alberta Parks Acknowledgments

http://www.albertaparks.ca/

40


41


The Problem

Inaccurate annual forecasts for passport demand caused funding imbalances among 40 passport offices

Volumes of 500 to 80,000/month across offices

National annual forecast for all offices with 25% variance

National forecast did not match the local office traffic

Customers were unhappy, delays were rampant

1.25 million passport queue!

42


The Outfall

Cost overruns

Staffing resource mismatches at local offices

Revenue stream impacted (transferring passports between offices cost $40/passport).

Customers were unhappy

43


The Fix

SAS® Forecast Server produced forecast for each division

Initial SAS-based forecast produced using a middle-out approach for each division

Division-wide forecast resulted in less variance within a couple of percent

But….they wanted better

44


Complicating forecast factors and their outfall

2007 U.S. law forced fliers to have a passport.

Calgary saw a huge spike in applications from airline passengers

2009 U.S. law required credentialed auto passengers

Alberta stayed flat because there are no „destination‟ cities across the border. “Nowhere to drive”

Eastern provinces saw a huge spike instead

Not all offices saw the same influx. Each office was different

Local office forecasting was required

45


The Next Level: Local Office SAS-based Forecasts

Accounted for complicating factors at local office level

Forecast then rolled up to national level with a resulting variance of 1% nationally and also by office!

Resource requirements much better anticipated

Processing time reduced from 1 month to 10 days

Forecasting work reduced from months to days

46


Observations

Victoria very pro-active with 6 month application lead time

St. John‟s and Quebec only 1 month application lead time

60% of Canadian citizens now have passports

All done with two people running SAS on a desktop machine

47


Urgent and express (0-9 days)

Renewal vs. non-renewal applications

Toddler, children and adult applications

Next Steps: More detailed factors to improve accuracy

48


Wisconsin Department of Revenue

Business Issue

Data spread out in multiple, isolated systems that slowed collection of tax dollars

Solution SAS® Enterprise Data Integration Server

SAS® Enterprise BI Server

SAS® Enterprise Miner™

Results/Benefits

Dramatic efficiency gains: reports that took 20 minutes can be done in 5 seconds

State brought in $32 million in revenue more quickly, and discovered $5 million that might have been lost

Hundreds of employees access information that once required a special request and custom-coded report

State & Local Government

Read the Full Story

“With this data warehouse,

we've been able to keep our

costs low and our productivity

high.”

Roger Ervin Secretary of Revenue

http://www.sas.com/software/data-management/entdiserver/index.html




http://www.sas.com/technologies/bi/entbiserver/index.html




http://www.sas.com/technologies/analytics/datamining/miner/




http://www.sas.com/success/wisconsindor.html

49


RCI Global Vacation Network

Business Issue

RCI needed a quick way to estimate the vale of time-share swaps and forecast demand.

Solution

SAS® Enterprise MinerTM and SAS®

Enterprise Guide ®

Results/Benefits

The company is saving $1 million a year by eliminating the need to maintain an unstable and inaccurate forecasting tool.

Services

Read the Full Story

“SAS is one of a kind because

it allows me to apply science

to the business of vacations,

which is quite amazing to me.”

Sri Raghavan Senior Vice President of Product

Development

http://www.sas.com/technologies/analytics/datamining/miner/index.html






http://www.sas.com/technologies/bi/query_reporting/guide/





http://www.sas.com/success/rci.html

50


SRA International, Inc.

Business Issue

Extract, analyze, categorize, index and apply metadata to a government organization‟s library with more than 100 years‟ worth of reports.

Solution

SAS® Enterprise Content Categorization

Results/Benefits

Re-indexed collection in hours rather than months or years.

Categorized accurately 90 percent of the time vs. 75 percent for human indexers.

Government

“SAS allows you to work under

the hood. It‟s not a black box.‟‟

Bill McKinney Taxonomy Manager

Read the Full Story

http://www.sas.com/text-analytics/enterprise-content-categorization/index.html




http://www.sas.com/success/sra.html

51


Telefónica O2 Germany Business Issue

Slow, inaccurate, decentralized forecasting

Difficulty supporting business planning, decisions

Solution

SAS® Forecast Server

SAS® Enterprise Guide

SAS® BI Server

SAS® DI Server

Results/Benefits

Forecasts improved by 18 percent, resulting in more accurate reports in areas like product and call center utilization, tariff migrations trends, contract extensions, terminations and churn

5

1


Communications

Read the Full Story

“The switch to SAS has proven to be

a strategic step, enabling the

company to expand its analytical

expertise and substantially advance

its business planning and

management.”

Markus Heimann

Vice President, Business Intelligence

http://www.sas.com/technologies/analytics/forecasting/forecastserver/index.html


http://www.sas.com/technologies/bi/entbiserver/

http://www.sas.com/technologies/dw/index.html



http://www.sas.com/success/telefonicaO2.html

52


Avantas Healthcare

Read the Full Story

“SAS has the ability to

incorporate hundreds of

different models, allowing us

to create a custom forecasting

methodology. That is very

important because every unit

in a hospital behaves so

differently."

Julie Kiefer Manager of Analytics

Business Issue

Improve forecasting capabilities

Develop more accurate forecasts of hospital labor needs

Deliver forecasts much more quickly than before

Solution

SAS® Forecast Server

Results/Benefits

Eight hours versus 80 hours to forecast a hospital system‟s labor needs

A 13 percent increase in forecast accuracy

Grow business fivefold without adding staff

Reduce time nurse managers spend on staffing issues

http://www.sas.com/success/avantas.html

http://www.sas.com/success/avantas.html





53


Bank of America

Business Issue

Outsourcing the development of OLAP cubes to another department to fill client forecast requests was time-consuming and inefficient

Solution

SAS® 9.2 and the SAS® Business Analytics Framework, specifically SAS® Enterprise BI Server and SAS® Enterprise Guide®

Results/Benefits

Building forecasts in one environment reduces processing time from months to days or hours, improves quality control, and gives forecasters more time to analyze data and build dashboards to meet client needs

Banking

“SAS … has reduced my

monthly reporting duties from

three days to three hours.”

Jessica Deeter Vice President, Consumer Product

Strategic Analyst

Read the Full Story

http://www.sas.com/software/sas9/




http://www.sas.com/businessanalytics/index.html












http://www.sas.com/success/bankofamerica.html

54


ODEC

Business Issue

Forecast Power Supply Needs for more than 1 Million Member Customers

Solution

SAS® Analytics with SAS® Forecast Server

Results/Benefits

Saved Utility Customers Millions of Dollars in its First Year Using SAS

Reduced Rates Four Times in a Single Year

Energy and Utilities

Read the Full Story

“We actually lowered the rate

we charge for wholesale

power four times in the past

year.”

David Hamilton Manager of Load Forecasting

http://www.sas.com/technologies/analytics/index.html







http://www.sas.com/success/odec.html

55


NOVEC

Business Issues

Forecast demand for power

Anticipate need for new substations or upgrades

Solutions

SAS® Analytics Pro, SAS® Enterprise Guide and SAS® ETS

Results/Benefits

46 percent less error in a SAS energy forecasting model compared with a naïve model

SAS model improves on other modeling software by 15 percent

Energy/Utilities

Read the Full Story

“SAS has the functionality to

do what we need now and

what we anticipate needing in

the future.”

Jamie Hall

Senior Operations Research Analyst

http://www.sas.com/technologies/analytics/statistics/analytics-pro/index.html





http://www.sas.com/technologies/analytics/forecasting/ets/

http://www.sas.com/success/novec.html

56


Orange Business Services

Business Issue

Enhance and maintain strong customer relationships

Grow business and maintain competitive edge

Understand, anticipate client needs and expectations

Solution

SAS® Customer Intelligence, SAS® for predictive analytics and data mining

Results/Benefits

Improved productivity by 30 percent

Met targets for responsiveness, pro-activity

Unified CRM strategy

Communications

“With productivity up by more

than 30 percent and the

achievement of optimal

information freshness, our

CRM solution has enabled us

to reach our targets for

responsiveness and pro-

activity in relation to our

market.”

Gaëlle Vallée Manager, Operational Data Mining Team

Read the Full Story

http://www.sas.com/software/customer-intelligence/index.html




http://www.sas.com/technologies/analytics/datamining/index.html





http://www.sas.com/success/orange-business-services.html

57


5

7


Insurance

Read the Full Story

“Since all our business

partners also use SAS, this

makes it even easier for

CIGNA to work with them,

because we can fully

integrate our processes for

database analytics and

product and channel

strategies.” Gary Denson

CEO, Cigna Insurance Thailand

CIGNA Thailand

Business Issue

The insurer needed to gain insights from data to boost marketing and partner relationships.

Solution

SAS® Analytics

Results/Benefits

Increased revenue.

Retail partner gained USD $600,000 in first three months.

Revenue gains paid for product in first month.

http://www.sas.com/success/cignathai.html




58


More information

Canada Post http://www.sas.com/offices/NA/canada/en/success/canada-post2011.html

http://www.sas.com/offices/NA/canada/en/success/canada-post2011.html





Business Analytics Forum - Sas Institute · “Percieved camper safety can be impacted by the level...

Documents

Transcript of Business Analytics Forum - Sas Institute · “Percieved camper safety can be impacted by the level...