Large data series: Modeling the usual to identify the unusual
{open_quotes}Standard{close_quotes} approaches such as regression analysis, Fourier analysis, Box-Jenkins procedure, et al., which handle a data series as a whole, are not useful for very large data sets for at least two reasons. First, even with computer hardware available today, including parallel processors and storage devices, there are no effective means for manipulating and analyzing gigabyte, or larger, data files. Second, in general it can not be assumed that a very large data set is {open_quotes}stable{close_quotes} by the usual measures, like homogeneity, stationarity, and ergodicity, that standard analysis techniques require. Both reasons dictate the necessity to use {open_quotes}local{close_quotes} data analysis methods whereby the data is segmented and ordered, where order leads to a sense of {open_quotes}neighbor,{close_quotes} and then analyzed segment by segment. The idea of local data analysis is central to the study reported here.
- Research Organization:
- Oak Ridge National Lab. (ORNL), Oak Ridge, TN (United States)
- Sponsoring Organization:
- USDOE Office of Energy Research, Washington, DC (United States)
- DOE Contract Number:
- AC05-96OR22464
- OSTI ID:
- 465877
- Report Number(s):
- CONF-970241-2; ON: DE97005119; TRN: 97:002777
- Resource Relation:
- Conference: 2. International Association for Statistical Computing world conference (IASC), Pasadena, CA (United States), 19 Feb 1997; Other Information: PBD: 1997
- Country of Publication:
- United States
- Language:
- English
Similar Records
Input/output modeling for process control
The systematic grouping of building sets and selection of building prototypes: BUCS as a pre-processor to Feds