Login

Publications  •  Project Statistics

Glossary  •  Schools  •  Disciplines
People Search: 
   
Title/Abstract Search: 

Dissertation Information for Stuart Ira Bretschneider

NAME:
- Stuart Ira Bretschneider

DEGREE:
- Ph.D.

DISCIPLINE:
- Business

SCHOOL:
- The Ohio State University (USA) (1980)

ADVISORS:
- None

COMMITTEE MEMBERS:
- None

MPACT Status: Incomplete - Not_Inspected

Title: ADAPTIVE TIME SERIES ANALYSIS AND FORECASTING

Abstract: "This dissertation presents a stream of research which has the purpose of integrating dynamic system modeling using time-varying parameters with forecasting for public sector application. Chapter I considers the problem of univariate forecasting, that is forecasting a time series using only time and the data series. A simple distributed lag model with time-varying parameters is specified and estimated using the Carbone-Longini Adaptive Estimation Procedure (AEP). Several aspects of AEP in this setting are examined. Relationships between forecast horizon, the length of a forecast into the future, and filter design are formulated and tested in comparative studies with other univariate forecasting techniques.

The attempt to evaluate forecast performance in the first chapter focused attention to the problem of validating or evaluating forecasts. The second chapter addresses the validation problem in a general fashion. Two post hoc experimental designs are formulated for distinct forecast objectives and the effects of different measures of accuracy are considered. Concepts and hypotheses developed in the chapter are illustrated with empirical results from a ""forecasting tournament"" where experienced users of several forecasting approaches independently developed forecasts of the same real time series data. The objective of this chapter is to provide prescriptive propositions about the design of a forecast evaluation study.

Multivariate forecasting models take advantage of additional information in the form of independent variables. Chapter III begins to consider how the adaptive forecasting approach may be extended to the multivariate setting. Relying on a model structure suggested by marketing theory, AEP is applied to estimation of parameters for the market response model. Forecasting peformance of this model is compared with other econometric and Box-Jenkins transfer function models for a specific data set. Alternative estimation procedures are considered in this context.

The work begun in Chapter III is extended in Chapter IV, to develop a general approach of multivariate adaptive forecasting. An empirical model identification procedure is developed to produce a forecast model for empirical data and parameters are estimated using the AEP technique. A procedure for accounting for future structural changes is incorporated into the forecasting procedure. Simulated and real data are used to evaluate the forecast performance of the adaptive approach with a naive no change forecast, univariate AEP approach, and the Box-Jenkins transfer function technique. The last chapter of the dissertation summarizes the results of this research and attempts to develop directions for future work."

MPACT Scores for Stuart Ira Bretschneider

A = 0
C = 2
A+C = 2
T = 0
G = 0
W = 0
TD = 0
TA = 0
calculated 2008-01-31 06:11:59

Advisors and Advisees Graph