Accession Number : ADA293929

Title :   Dynamic Linear Models with Leading Indicators. Revision.

Descriptive Note : Master's thesis,

Corporate Author : GEORGE WASHINGTON UNIV WASHINGTON DC

Personal Author(s) : Chen, Jingxian

PDF Url : ADA293929

Report Date : DEC 1991

Pagination or Media Count : 46

Abstract : This thesis proposes a dynamic linear model (DLM) to deal with the problem of forecasting with leading indicators. We call this type of a DLM as a dynamic linear model with leading indicators. Our approach expands the conventional one-dimension DLMs to the two dimension case. Analyses of some real data sets which initially motivated us to explore our approach, are used as applications. For reasons of confidentiality they have been coded as Data Set One, Data Set Two and Data Set Three, respectively. Our approach has a much wider field of application, for instances, the two-dimension filter problems in image processing, and estimation problems related to Markov random fields.

Descriptors :   *MATHEMATICAL MODELS, *FORECASTING, *LINEARITY, DATA BASES, LINEAR SYSTEMS, IMAGE PROCESSING, SIZES(DIMENSIONS), DYNAMICS, TWO DIMENSIONAL, TIME SERIES ANALYSIS, THESES, PROBLEM SOLVING, ESTIMATES, FILTERS, INDICATORS, MARKOV PROCESSES.

Subject Categories : Operations Research

Distribution Statement : APPROVED FOR PUBLIC RELEASE