预测与时间序列（英文版·第3版）

    书籍
    数学书籍
本书是预测与时间序列分析课程的教材，书中讲解了预测的重要过程以及可以用于预测的各种统计技术。作者清晰地展示了在营销、金融，人力资源管理，产品调度，过程控制和策略管理中通过预测做出明智决策的重要性。
本书适合作为工商管理、理工(包括数学、统计学、计算机科学等)类高年级本科生和研究生的教材，同时可以作为需要进行现实预测的专业人员的参考书。
本书的主要特点
清晰、完善地介绍了Box-Jenkins方法精确、易于理解地讨论了传递函数和干涉模型，并介绍了多元时间序列分析给出了基于真实案例的大量习题使用MINITAB和SAS输出给出预测的结果，并有选修的章节详细讲述MINITAB和SAS的用法

CONTENTS
PART I
INTRODUCTION
CHAPTER 1
AN INTRODUCTION TO FORECASTING 2
1.1 Introduction
1.2 Forecasting and Time Series
1.3 Forecasting Methods
1.4 Errors in Forecasting
1.5 Choosing a Forecasting Technique
1.6 An Overview of Quantitative Forecasting Techniques
1.7 Computer Packages: Minitab and SAS
Exercises

CHAPTER 2
BASIC STATISTICAL CONCEPTS
2.1 Populations
2.2 Probability
2.3 Random Samples and Sample Statistics
2.4 Continuous Probability Distributions

　　 Forecasting and Time Series: An Applied Approach, third edition, is designed as a text-book for applied courses in forecasting and time series analysis and as a reference book for practitioners who must make real world forecasts. It is appropriate for advanced (junior-and senior-level) undergraduates and graduate students in business, engineering, and the sciences (including mathematics, statistics, operations research, and computer science). The required mathematical and statistical background for this book is college algebra and basic statistics.
　　 This third edition attempts to combine and somewhat expand the best aspects of the first two editions. The first edition began with a short discussion of regression analysis,proceeded to a complete discussion of exponential smoothing and time series decompo-sition techniques, and concluded with a moderately complete presentation of the Box-Jenkins methodology. The second edition began with a more complete treatment of the Box-Jenkins methodology. This methodology was then used to integrate and unify most of the remaining forecasting techniques presented. We have found that some instructors who liked the first edition did not like the organization and Box-Jenkins emphasis of the second. On the other hand, many instructors preferred these aspects of the second edition. The third edition is organized into five parts in a way that should satisfy both those who prefer the first edition and those who prefer the second. Some instructors may wish to teach a course that covers all five sections. However, most courses will not cover
　　the entire text. For instance, some instructors will wish to build their course around the
　　regression/exponential smoothing/decomposition material. Others will wish to build their course around the Box-Jenkins methodology. We now describe the five parts and then explain how they can be structured into different courses.
　　 Part I consists of an introduction to forecasting (Chapter 1) and a review of basic statistical concepts (Chapter 2). Part Il discusses forecasting by using regression analysis.
　　This part begins with Chapter 3, which presents simple linear regression. Chapter 4 discusses multiple regression, including an introduction to model building and residual analysis. Chapter 5 presents various advanced topics in regression. These topics are not needed for studying any other chapter in the book, but would be covered by an instruc-tor interested in a more complete discussion of using regression analysis in forecasting.Part III discusses forecasting by using time series regression, decomposition methods,and exponential smoothing. The prerequisite for reading this part is a basic knowl-edge of simple and multiple regression analysis, as provided by Chapters 3 and 4 ofPart II.
　　 We suspect that some forecasting courses will have a regression prerequisite, and instructors teaching such courses will begin with Part III. Therefore, we have written this material to stand on its own as much as possible. Part III begins with Chapter 6, which covers time series regression. This includes modeling trends and seasonal effects by using polynomial functions of time, dummy variables, and trigonometric functions.Also covered is an introduction to modeling autocorrelated error terms. Chapter 7 dis-cusses time series decomposition methods. Part III concludes with Chapter 8, which pre-sents exponential smoothing. Included are discussions of simple and double exponential smoothing, Winters' Method, and damped trend methods.
　　 Parts IV and V discuss forecasting by using the Box-Jenkins methodology. These parts are written from first principles and can be read without reading Part II or Part III. Therefore, an instructor may begin a course with Part IV, which presents basic techniques of the Box-Jenkins methodology. Chapters 9 and 10 begin this part and discuss nonseasonal Box-Jenkins modeling. Chapter 11 concludes this part and presents an introduction to Box-Jenkins seasonal modeling. It is important to note that, in order to simplify notation, we have delayed use of the backshift operator until Part V. There-fore, the reader can obtain from Part IV a complete knowledge of nonseasonal Box-Jenkins modeling and an introduction to seasonal Box-Jenkins modeling without using this operator. Part V begins with Chapter 12, which presents a more advanced treatment
　　of seasonal Box-Jenkins modeling. Chapter 13 covers the use of the Box-Jenkins meth-odology in time series regression and exponential smoothing. This chapter refers the reader to the necessary prerequisite parts of Chapters 6 and 8 as needed. Part V concludes with Chapter 14, which discusses transfer functions and intervention models.
　　 Below we list some possible courses that can be based on this book. All courses are
　　 assumed to include the introduction to forecasting provided in Chapter 1 and any needed basic statistical review from Chapter 2.
　　 1. A course on forecasting by using regression analysis, time series regression, decomposition methods, and exponential smoothing would consist of Parts II and III. A more intensive course would also include the basic techniques of the Box-Jenkins methodology, as given by Part IV.
　　 2. A course on forecasting by using time series regression, decomposition methods, exponential smoothing, and the basic techniques of the Box-Jenkins methodology would consist of Parts III and IV. A more intensive course would also include the advanced techniques of the Box-Jenkins methodology, as given by Part V.
　　 3. A course on forecasting by using the Box-Jenkins methodology would consist of Parts IV and V, with relevant portions of Chapters 6 and 8 included as indicated by the discussions of Chapter 13.
　　 4. A course on forecasting by using regression analysis and the Box-Jenkins methodology would consist of Part II, Chapter 6 of Part III, and Part IV. A more intensive course would also include portions or all of Part V. We have placed a premium in this book on illustrating forecasting by using many real world data sets in the examples and exercises. In addition, we utilize Minitab and SAS outputs to present forecasting results and show in optional sections how to use these packages. A data disk that contains many of the time series in this book is also available.
　　 We wish to thank Kathleen Billus, Marcia Cole, Curt Hinrichs, Susan London,Michael Payne, and the other fine people at Duxbury Press for their help in this writing endeavor, as well as Rachel Youngman of Hockett Editorial Service. We would also like to thank the reviewers of this book. We would especially like to thank S. Chakraborti,University of Alabama; Terry Dielman, Texas Christian University; Benito Flores, Texas A & M University; Michael L. Hand, Willamette University; Robert McAuliffe, Babson College; Helmut Schneider, Louisiana State University; Stanley R. Schultz, Cleveland State University; and Mack C. Shelley, II, Iowa State University for their many useful comments and suggestions. Finally, we thank our wives and children for their love and encouragement.
　　 Bruce L. Bowerman
　　 Richard T. O'Connell