Bayesian Forecasting and Dynamic Models (Springer Series in Statistics) (Hardcover)
Mike West (Author), Jeff Harrison (Author)  


Editorial Reviews
Product Description
The second edition of this book includes revised, updated, and additional material on the structure, theory, and application of classes of dynamic models in Bayesian time series analysis and forecasting. In addition to wide ranging updates to central material in the first edition, the second edition includes many more exercises and covers new topics at the research and application frontiers of Bayesian forecastings.


Product Details

• Hardcover: 700 pages
• Publisher: Springer; 2nd edition (January 24, 1997)
• Language: English
• ISBN-10: 0387947256
• ISBN-13: 978-0387947259

CONTENTS
Preface v
1 Introduction 1
1.1 Modelling, Learning and Forecasting 1
1.2 Forecast and Decision Systems 8
1.3 Bayesian Modelling and Forecasting 20
1.4 Historical Perspective and Bibliographic Comments 28
2 Introduction to the DLM 32
2.1 Introduction 32
2.2 The DLM and Recurrence Relationships 34
2.3 The Constant Model 39
2.4 Specification of Evolution Variance Wt 49
2.5 Unknown Observational Variances 52
2.6 Illustration 57
2.7 Appendix 61
2.8 Exercises 61
3 Introduction to the DLM 68
3.1 Introduction 68
3.2 The Multiple Regression DLM 73
3.3 Dynamic Straight Line through the Origin 74
3.4 Model Variances and Summary 83
3.5 Exercises 91
4 The Dynamic Linear Model 97
4.1 Overview 97
4.2 Definitions and Notation 100
4.3 Updating Equations: The Univariate DLM 103
4.4 Forecast Distributions 106
4.5 Observational Variances 108
4.6 Summary 111
4.7 Filtering Recurrences 112
4.8 Retrospective Analysis 116
4.9 Linear Bayes’ Optimality 122
4.10 Reference Analysis of the DLM 128
4.11 Appendix: Conditional Independence 136
4.12 Exercises 138
5 Univariate Time Series DLM Theory 143
5.1 Univariate Time Series DLMS 143
5.2 Observability 143
5.3 Similar and Equivalent Models 148
5.4 Canonical Models 154
5.5 Limiting Results for Constant Models 160
5.6 Stationarity 169
5.7 Exercises 172
6 Model Specification and Design 178
6.1 Basic Forecast Functions 178
6.2 Specification of Ft and Gt 186
6.3 Discount Factors and Component Model Specification 193
6.4 Further Comments on Discount Models 200
6.5 Exercises 202
7 Polynomial Trend Models 208
7.1 Introduction 208
7.2 Second-Order Polynomial Models 211
7.3 Linear Growth Models 217
7.4 Third-Order Polynomial Models 225
7.5 Exercises 229
8 Seasonal Models 234
8.1 Introduction 234
8.2 Seasonal Factor Representations of Cyclical Functions 237
8.3 Form-Free Seasonal Factor DLMS 238
8.4 Form-Free Seasonal Effects DLMS 240
8.5 Trend/Form-Free Seasonal Effects DLMS 244
8.6 Fourier Form Representation of Seasonality 246
8.7 Exercises 263
9 Regression, Autoregression, and Related Models 270
9.1 Introduction 270
9.2 The Multiple Regression DLM 270
9.3 Transfer Functions 281
9.4 Arma Models 291
9.5 Time Series Decompositions 301
9.6 Time-Varying Autoregressive DLMS 304
9.7 Exercises 311
10 Illustrations and Extensions of Standard DLMS 317
10.1 Introduction 317
10.2 Basic Analysis: A Trend/Seasonal DLM 318
10.3 A Trend/Seasonal/Regression DLM 330
10.4 Error Analysis 345
10.5 Data Modifications and Irregularities 350
10.6 Data Transformations 353
10.7 Modelling Variance Laws 357
10.8 Stochastic Changes in Variance 359
10.9 Exercises 365
11 Intervention and Monitoring 369
11.1 Introduction 369
11.2 Modes of Feed-Forward Intervention 374
11.3 Illustrations 383
11.4 Model Monitoring 392
11.5 Feed-Back Intervention 401
11.6 A Bayes’ Decision Approach to Model Monitoring 415
11.7 Exercises 423
12 Multi-Process Models 427
12.1 Introduction 427
12.2 Multi-Process Models: Class I 429
12.3 Multi-Process Models: Class II 443
12.4 Class II Mixture Models Illustrated 456
12.5 Exercises 486
13 Non-Linear Dynamic Models: Analytic and Numerical Approximations 489
13.1 Introduction 489
13.2 Linearisation and Related Techniques 492
13.3 Constant Parameter Non-Linearities: Multi-Process Models 497
13.4 Constant Parameter Non-Linearities: Efficient Numerical Integration 499
13.5 Efficient Integration for General Dynamic Models 505
13.6 A First Simulation Method: Adaptive Importance Sampling 509
14 Exponential Family Dynamic Models 516
14.1 Introduction 516
14.2 Exponential Family Models 517
14.3 Dynamic Generalised Linear Models 521
14.4 Case Study in Advertising Awareness 534
14.5 Further Comments and Extensions 555
14.5 Exercises 558
15 Simulation-Based Methods in Dynamic Models 561
15.2 Introduction 561
15.2 Basic MCMC in Dynamic Models 565
15.3 State-Space Autoregression 571
16 Multivariate Modelling and Forecasting 581
16.1 Introduction 581
16.2 The General Multivariate DLM 582
16.3 Aggregate Forecasting 586
16.4 Matrix Normal DLMS 597
16.5 Other Models and Related Work 626
16.6 Exercises 627
17 Distribution Theory and Linear Algebra 631
17.1 Distribution Theory 631
17.2 The Multivariate Normal Distribution 636
17.3 Joint Normal/Gamma Distributions 640
17.4 Elements of Matrix Algebra 645
Bibliography 652
Author Index 667
Subject Index 671