George Mason University
CSI/Statistics Colloquium Series
Seminar Announcement

Analysis of Superlarge Industrial Datasets

David Banks

National Institute for Standards and Technology


ABSTRACT

It is common to encounter superlarge, highly multivariate data in industrial applications. Such data pose four special problems:

  • compression
  • preanalysis
  • analysis
  • indexing This talk concentrates on the second and third problems, using examples from PPG and semiconductor manufacture. The key points will be to review a protocol for preparing a superlarge dataset for analysis, to describe a comparative survey of new-wave nonparametric techniques that apply to such data, and to urge the estimation of local dimensionality as a tool in these analyses. Additionally, there will be a brief discussion of the hard question of index creation, with some comments about possible solution strategies.


    Friday, March 5, 1999
    George W. Johnson Center, Assembly Room D
    Seminar at 10:45 a.m.
    Refreshments at 10:30 a.m.


    Information about the Statistics Colloquium Series, including directions, and current and past schedules, is available at www.science.gmu.edu/statseminars