Customer Reviews

Introduction to Time Series and Forecasting

5 star:		(5)
4 star:		(4)
3 star:		(1)
2 star:		(4)
1 star:		(1)

 

 

Very good introductory book to ARMA models. Full of real-life examples that provide some intuitive insight about the issues that may arise when modelling time series and forecasting. Requires some initial knowledge in statistics and algebra but if you're involved in time series modelling, it should be your first book.

In contrast to their graduate text "Time Series: Theory and Methods" this book is more elementary and introductory and is pitched at the advanced undergraduate level requiring only calculus, elementary statistics and matrix algebra. It gives very good coverage to a wide variety of time series models and includes some nonstationary models. In this second edition the chapter on nonstationary models includes the latest coverage of ARCH and GARCH models presented in a way that I found very accessible.
Computations are done with ITSM and in this edition the ITSM 2000 version 7.0 edition is included on a CD so that students can reproduce the authors' calculations and run analyses of their own.

Another nice feature of the text that distinguishes it from other texts at this level is the introduction of multivariate time series, coverage of state space models, chaos and cointegration. Ideas are illustrated with examples. Important theory is discussed but is kept brief and theorems and proofs are not given to the extent of their other more theoretical text.

First off, my background is as follows: I have taken master's level courses in probability, statistical inference, linear regression, linear algebra, and differential equations from a very reputable university, and received high marks in all classes. Also, I've scored 800 on the Quantitative section of the GRE.

So you would think that a person like myself would find this book to be challenging yet rewarding and manageable. After spending 2 weeks trying to decipher the text, and going over 3 chapters, I've lost all hope in the book. When I read this textbook, it feels as though I am trying to understand the Authors' stream of consciousness.

As a student of math, I am quite used to spending countless hours deciphering the theorems and propositions of many mathematicians. But with this book, I also have the added burden of having to dicipher the Authors' thought process. To be clear, the material in the text book is very good, but the presentation of the material is that of a "rambler." At this point I have given up on the book as a source of my learning, and have purchased another text book to use as my reference.

I agree with the many other reviewers who stated that the book is unorganized and written poorly. It seems as though they spent MINIMAL time on producing this book just to meet a deadline. After the first few chapters, it REALLY gets annoying and makes you want to chastise the Authors for being so irresponsible. They may be geniuses in their field, but they have no right to teach the material with these kinds of products.

*UPDATE* 11DEC2010
I've more or less gone through the entire book(only by necessity since the homework questions were from the text). My opinion of this textbook still stands.
As an alternative I recommend "Analysis of Financial Time Series" by Tsay. I've read bits of it, and it seems very well written and progressive. Helped me out a TON in understanding ARCH/GARCH processes.

I think the book is not written in a very organized way. It's not a book for picking up time series quickly. It's saturated with information, which I'm not sure if it's necessary for implementation. I have no problem following the math, however, if I want to pick up something and implement it within a day or two, the book is a bit harder to digest. Wouldn't think this is an undergraduate course book as it covers convergence in probability or mean-squared, which I learnt in PhD courses, not even master level.

I'm sure it's a great book for people with advanced mathematics backgrounds, but it's simply NOT an introductory-level text, and it's unsuitable for engineers and other scientists. Definitely not an undergraduate-level text (although I should mention I'm trying to use it for my graduate-level coursework.) I wish I could rate the actual content of the book, but it's too incomprehensible for me to give an informed appraisal. Suffice it to say that the book's title misrepresents the content. THIS IS NOT AN INTRODUCTION.

Apart from that, the exercise problems have no solutions given, and Springer (the book's publisher) refuses to let me (a graduate student) have access to the solutions manual. I'd have to be teaching a class using the book, for which I'd need to understand the material, for which a solutions manual would help - you get the idea.

Very good introductory book to ARMA models. Full of real-life examples that provide some intuitive insight about the issues that may arise when modelling time series and forecasting. Requires some initial knowledge in statistics and algebra but if you're involved in time series modelling, it should be your first book.

This does not appear to be a true "introductory" text. I would not recommend it as a first text on time series.
A good bit of the text is spent walking through derivations of formulae, and many of the exercises do not use actual data, but are directions such as "Derive the recursions for the Burg estimates..." or "Derive a cubic equation for the maximum likelhood estimates of..." There is too little information on how to proceed in analyzing actual data.
The book is not well-organized, but appears to have been thrown together hurriedly, and lacks any connecting flow. Coverage of topics is more brief as you get to the later chapters, which is another indication of rushing something into print.

A decent basic introduction covering a lot of topics. It's much more accessible for learning the subject for the first time then many other books which pile on the mathematical notation and obscure the actual meaning of things. The accompanying CD is very nice, although it gets annoying very fast that you're restricted to very small dataset sizes---but it does help in learning. The only two things that are somewhat of a problem with this book are 1) many times, rather than clearly stating "here's the algorithm you need to implement", you are referred to 3 or 4 other sections of the book for pieces of the algorithm, often without a clear explanation of exactly how that earlier section is supposed to be worked into the current desired algorithm and 2) there aren't a lot of practical insights as to how to actually initialize many of the algorithms (everything is great if you already know all the parameters in advance but starting from scratch with just raw data isn't dealt with I think as fully as would be useful). All in all, though, the book is helpful and, as I said, very good for learning the essential concepts for the first time.

This is one of those books that you can't find much cons to it. The book is inexpensive, and it's unbelievably lightweight. The material is rich, and yet easy to understand. The author actually brings you step by step from elementary to theorectical proofs.

I am kind of confused now, the book says there is a CD included with the book, but my book came alone without any CD? Did the seller make a mistake or it has other way to access to the CD?