DMS 9610157 WORKSHOP ON STOCHASTIC MODEL BUILDING AND VARIABLE SELECTION ABSTRACT As statistical analysis and modeling of larger and more complex data sets is becoming more common, new and better computer based tools for guiding the initial specification of the main features of statistical models are being developed. Recently, a new generation of stochastic algorithms has been emerging as an important augmentation to traditional deterministic strategies. Examples include stochastic search methods for variable selection, graphical models, selection of variable transformations and interactions, wavelet thresholding, ARMA modeling, CART, MARS, neural networks and more. A workshop on stochastic model building and variable selection that blends methodology and case studies will significantly advance this field of statistical research and will encourage dissemination of an important class of modeling techniques to practitioners. The workshop is organized in coordination with the NBER/NSF Time Series Seminar and with the Workshop on Wavelets in Statistics. The common research themes between stochastic model building, wavelets analysis and times series analysis will be exploited to create both a synergy at the research level and savings in travel expenses and arrangements. Funds from this grant will be used for domestic travel expenses for participants based in the U.S.A, particularly junior researchers, and others from underrepresented groups.

Agency
National Science Foundation (NSF)
Institute
Division of Mathematical Sciences (DMS)
Type
Standard Grant (Standard)
Application #
9610157
Program Officer
James E. Gentle
Project Start
Project End
Budget Start
1997-09-01
Budget End
1998-08-31
Support Year
Fiscal Year
1996
Total Cost
$9,000
Indirect Cost
Name
Duke University
Department
Type
DUNS #
City
Durham
State
NC
Country
United States
Zip Code
27705