Transcript Case Studies in Data Mining at the Keck School

Is There A Doctor In The House ?
Case Studies in Data Mining at the
Keck School of Medicine of USC
Presented by
Michael Seymour
Senior Director, Development Operations
Keck School of Medicine / University of Southern California
CARA Seminar Day
October 21, 2005
Biola University
The Theory Behind the Madness
• Data Mining –
-- needs to be practical (not just doing it for fun or to
be a statistician)
-- needs to be creative (think outside the box and play around)
look for: consistent patterns, trends, irregularities or anomalies
-- the searches you perform need to ask and answers questions that help in the overall
development process
-- needs to produce results
The aim of data mining is to extract implicit previously unknown and
potentially useful or actionable patterns and model from data.
The Theory Behind the Madness, Continued
•
Data Mining empowers users to direct their activities by delivering accurate
and useful information not available in raw data.
•
Decisions in these situations are based on sound database intelligence, not
on instinct or emotions.
MODELS
Clustering: identify groups of items that share a particular characteristic which
is not pre-defined.
Association: identify relationships between events that occur at one time.
Sequencing: similar to “Association” except that the relationship exists over a
period of time.
Actual cases (real life examples)
• CASE # 1
Alumni Annual Fund
• CASE # 2
Cancer Direct Mail Program
•
KDnuggets : Polls
Successful Data Mining Applications (July 2005) Poll Industries/fields where you successfully
applied data mining in the past 3 years [149 replies, 421 votes total]
Banking (51) 12%
Biotech/Genomics (11) 3%
Credit Scoring (35) 8%
CRM (52) 12%
Direct Marketing/ Fundraising (34) 8%
e-Commerce (11) 3%
Entertainment/ Music (4) 1%
Fraud Detection (31) 7%
Gambling (2) 0%
Government applications (12) 3%
Insurance (24) 6%
Investment / Stocks (5) 1%
Junk email / Anti-spam (5) 1%
Health care/ HR (15) 4%
Manufacturing (19) 5%
Medical/ Pharmaceutical (12) 3%
Retail (25) 6%Science (17) 4%
Security / Anti-terrorism (5) 1%
Telecom (23) 5%Travel/Hospitality (8) 2%
Web (9) 2%
Other (11) 3%
Case #1 Alumni Annual Fund
Background:
The Class of 1957 is considered one of our
more active classes.
What we did:
In FY 2004, the Class of 1957 donors with previous giving of $1,000+
were targeted for a solicitation ask of $2,500+. We used our P!N data
to identify alumni within the class and mailed a targeted solicitation
letter asking for a certain amount of money, based on the P!N data
and their previous giving. The ask for specifically for the class
of 1957 scholarship fund.
Data / Results
FY:
# mailed to:
# of gifts:
total amount:
2006
51
8
$8,600
2005
51
23
$24,600
[Note: 2006 data is only 3 months into fiscal year]
2004
52
30
$31,840
2003
53
22
$12,704
Case # 1 Alumni Annual Fund, continued
• Analysis:
FY 2003
7 $1,000 + gifts
6 $100 + gifts
FY 2004
2 $10,000 gifts !
$19,135.12 increase from previous year
FY 2005
decrease of $7,239.99
FY 2006
1 $5,000 gift to date
Case # 2 Cancer Direct Mail Program
•
Background:
Annual direct mail program for Cancer Center
involving acquisition and renewals
•
Plan:
The idea was to move $50 / $100 / $250 to $500 / $1,000
donors. In FY 2004, we used P!N data to look at Cancer
Direct Mail donor pool to move these donors to Cancer
Research Associates (CRA) membership.
Data / Results
FY:
2006
2005
Total acquisition
pieces mailed
525,000
525,000
599,288
605,243
Total renewal
pieces mailed
259,500
253,618
284,968
301,512
# gifts:
1,978
26,552
28,874
31,593
$1,299,467
$1,313,182
$1,353,450
total amount:
$83,017
(only 3 month
into fiscal year)
2004
2003
Case #2 Cancer Direct Mail Program,
Continued
• Analysis:
FY 2003
$25,000
$10,000
Total $ raised: $239,561
Total donors: 208
2
0
FY 2004
$25,000
$10,000
$5,000
$2,500
7
10
$1,000
$500
28
19
$1,000
$500
21
17
Total $ raised: $157,739
Total donors: 166
1
1
$5,000
$2,500
7
10
Case #2 Cancer Direct Mail Program,
Continued
Analysis, Continued
FY 2005
$25,000
$10,000
Total $ raised: $97,191
Total donors: 147
$1,000
$500
22
18
$25,000
0
$5,000
0
$1,000
$10,000
0
$2,500
2
$500
[Note: 2006 data is only 4 months into fiscal year]
4
0
FY 2006
0
0
$5,000
$2,500
4
18
Total $ raised: $10,889
Total donors: 32
Sources
• “Deep Pockets – Where the Alumni Money Is” by Peter B. Wylie
• “The Many Facets of Data Mining” by Peter B. Wylie
• “Data Mining for Fund Raisers” by Peter B. Wylie
• QUESTIONS AND ANSWERS
• COMMENTS
• DISCUSSION
Contact Information
Michael D. Seymour
Senior Director, Development Operations
Keck School of Medicine
University of Southern California
(323) 442 – 3890
[email protected]