Data Mining - Opportunity and Challange

Download Report

Transcript Data Mining - Opportunity and Challange

Data Mining - Opportunity and
Challenge
Chair - Al Holm
Scribe - Steve Beckwith
.ppt - Chuck Pullen
Fly on Wall - Lee Anne Willson
John Good
Bill Dillon
Concept Summary
 Organized data mining may offer a new way
for AAVSO to serve the professional
astronomy community while encouraging
new membership and new life for some
existing members.
 Some changes to existing data collection
and management would be needed.
 Entwined with entire concept of data
quality of visual observations.
The Promise of Data Mining
Activity for beginner
to advanced member
 Requires no telescope
or observations
 Analysis desperately
needed, in both AAVSO
and other data sets
 May appeal to new
types of “observers”

The Peril of Data Mining
Requires
infrastructure
 Infrastructure = $
 Requires Training
 Training = $
 Requires evaluation of
AAVSO data quality
 Evaluation = pain!

Concept Overview- Data
Enhancement
 Existing data base needs to be upgraded to
“validated”
 Rapid, machine-based screening of data
necessary to keep interest
 Some measure of data quality necessary,
on both visual, PEP and CCD data.
 Vol. archival of FITS CCD frames, with pro
level header data, important for NVO (more
later).
Concept Overview Infrastructure
 Cross platform, free to user, validated
software, from simple educational level to
high quality analytical .
 Multi level query
 Rapid access to new data, perhaps with
quality flags
 In depth training - quantum enhancement
of HOA
Concept Overview- Marketing
 Who are the miners?
 Recruit via the web, non-astronomy based
technology media, Sci. Amer., SETI@HOME
example
 Be inclusive of diverse levels, 4th graders
to PhDs, but be wary of results.
 Define goals and recognize work.
Concept Overview - Clients
 Who uses the output?
 Members
 Scientific Community
 Educators
 The curious...
What about the mass of
unevaluated survey data?
 Data analysis starts at home
 If we build it - they will come
 Perfect our system in house first, then
invite others to join us.
Potholes on the Info. Super Hwy.
 Interest?
 Funding (NVO funds possible)
 Sociopolitical
 Data flow constrictions (CCD Images)
 Automation of data quality evaluation
Recommendations
 Develop working group to consult.
 Combine efforts with overall NVO
participation.
 Update HOA to reflect new opportunities
 Prepare for change…
Conclusion
AAVSO’s mission to promote the
science of variable stars requires, in
addition to the collection of data,
using an integrated suite of database
design/upgrade, software, and training
for data analysis.
CCD/film Data Archival
Important future
contribution
 Well within goals of
AAVSO
 Separate but related
with data mining
 Address as part of NVO
