Home
Resource
Topics in compositional, seasonal and spatial-temporal time series
PDF
PDF format is widely accepted and good for printing.
Plug-in required
PDF-1(6.99 MB)
Citation & Export
View Usage Statistics
Staff View

Citation & Export
Hide

Simple citation

Chang, Kun. Topics in compositional, seasonal and spatial-temporal time series. Retrieved from https://doi.org/doi:10.7282/T3833V03

Export

Click here for information about Citation Management Tools at Rutgers.

Statistics
Hide

Description

TitleTopics in compositional, seasonal and spatial-temporal time series
NameChang, Kun (author); Chen, Rong (chair); Xie, Minge (internal member); Xiao, Han (internal member); Cheng, Jerry (outside member); Rutgers University; Graduate School - New Brunswick
Date Created2015
Other Date2015-10 (degree)
SubjectStatistics and Biostatistics, Time-series analysis, Prediction theory
Extent1 online resource (ix, 102 p. : ill.)
DescriptionThis dissertation studies several topics in time series modeling. The discussion on seasonal time series, compositional time series and spatial-temporal time series brings new insight to the existing methods. Innovative methodologies are developed for modeling and forecasting purposes. These topics are not isolated but to naturally support each other under rigorous discussions. A variety of real examples are presented from economics, social science and geoscience areas. The second chapter introduces a new class of seasonal time series models, treating the seasonality as a stable composition through time. With the objective of forecasting the sum of the next $ell$ observations, the concept of rolling season is adopted and a structure of rolling conditional distribution is formulated under the compositional time series framework. The probabilistic properties, the estimation and prediction, and the forecasting performance of the model are studied and demonstrated with simulation and real examples. The third chapter focuses on the discussion of compositional time series theories in multivariate models. It provides an idea to the modeling procedure of the multivariate time series that has sum constraints at each time. It also proposes a joint MLE method for threshold vector-error correction models. This chapter interprets the methodologies with an real example of the U.S. household consumption expenditures data. Threshold cointegration effects are analyzed on the U.S. real GDP growth rate. The estimation of TVECM is compared by the current two-step estimation method and the proposed joint MLE approach. Sensor allocation problem is studied in Chapter 4 under spatial-temporal models in Gaussian random fields. Given observations from existing sensors, the problem is solved by minimizing the integrated conditional variance based on different forecasting purposes. In this chapter, the time series are measured at different locations with both spatial and time series correlations. The spatial-temporal covariance structure is extensively discussed under both separable and nonseparable cases. The model is finally applied to the ozone level measurements in Harris County, Texas.
NotePh.D.
NoteIncludes bibliographical references
Noteby Kun Chang
Genretheses, ETD doctoral
Persistent URLhttps://doi.org/doi:10.7282/T3833V03
Languageeng
CollectionGraduate School - New Brunswick Electronic Theses and Dissertations
Organization NameRutgers, The State University of New Jersey
RightsThe author owns the copyright to this work.
Version 8.5.5
Rutgers University Libraries - Copyright ©2024