
Accession Number : ADA191157
Title : Partial Likelihood Analysis of Time Series Models, with Application to RainfallRunoff Data,
Corporate Author : MARYLAND UNIV COLLEGE PARK DEPT OF MATHEMATICS
Personal Author(s) : Slud, Eric V ; Kedem, Benjamin
PDF Url : ADA191157
Report Date : 25 Feb 1988
Pagination or Media Count : 25
Abstract : A general logisticautoregressive model for binary time series or longitudinal responses is presented, generalizing the discretetime Cox (1972) model with timedependent covariates as well as the recent regression models of Kaufmann (1987) for categorical timeseries. Since this model is formulated in terms of the timeseries covariates which are not themselves explicitly modelled, the largesample theory of parameterestimation must be justified by means of Partial Likelihood in the sense of Cox (1975), using theoretical results like those of Wong (1986). The largesample theory also justifies goodness of fit tests analogous to the chisquared tests of Schoenfeld (1980) and to the tests based on sums of (normalized) squared residuals used in logistic regression. These ideas are illustrated by analysis of a rainfallrunoff hydrological dataset previously analyzed by Yakowitz (1987).
Descriptors : *MATHEMATICAL MODELS, *TIME SERIES ANALYSIS, *APPLIED MATHEMATICS, *RAINFALL, *RUNOFF, LOGISTICS, REGRESSION ANALYSIS, HYDROLOGY, MAXIMUM LIKELIHOOD ESTIMATION, TIME DEPENDENCE
Subject Categories : Numerical Mathematics
Meteorology
Distribution Statement : APPROVED FOR PUBLIC RELEASE