Welcome to StatNotes.Net and MathNotes.Net!
As I have moved into research, I have heard a lot of book
recommendations. For now, I will collect them here. I'll add my
reviews on as I read them.
- Books about Time Series
- Brockwell and Davis's Time
Series: Theory and Methods: This
book is my favorite book on time series. It discusses both the
frequency domain and the time domain, it has a chapter on multivariate
time series, and it has a few sections at the end on long memory. I
have spent a long time with this book (especially the two aforementioned
chapters.)
- Priestley's
Spectral
Analysis and Time Series. Volumes I and II in 1 book. (Probability and
Mathematical Statistics): This book covers both univariate time
series (Volume 1) and multivariate time series (Volume 2). I believe it
provides the best coverage of cross-spectral densities and other
multivariate frequency domain concepts. (It does not cover long
memory time series, probably because the book predates them.) Sadly,
the
book
is
out of
print.
- Bloomfield's Fourier
Analysis of Time
Series: An Introduction (Wiley Series in Probability and Statistics): This book is one of the classics of time series analysis. It is not as thorough as
Priestley's book (above), but it does mention the analysis of multiple
time series.
- Koopmans's
The
Spectral Analysis of Time Series (Probability and Mathematical
Statistics): Yet another classic book about time series in the
frequency domain.
- Hamilton's Time Series
Analysis:
This was the first graduate-level econometrics book I encountered. It covers a large number of topics thoroughly, though it does ignore the frequency domain
entirely.
- Time
Series With Long Memory (Advanced Texts in Econometrics): This book,
by Peter Robsinon (one of the big names in long memory time series
research), provides a good review of long memory time series and
then reproduces a number of the "classic" papers in long memory time
sereis (by Robinson himself and by others).
- Multiple
Time Series (Wiley Series in Probability and Mathematical
Statistics): This book is by E. J. Hannan, the author of many time
series papers in the 1970's.
- Handbook
of Time Series Analysis: Recent Theoretical Developments and
Applications: This book is a compilation of articles about time
series. It first caught my interest because it included a recent
article by
Manfred Deistler about signal extraction and factor analysis. I
really like the idea of handbooks that summarize the latest
knowledge in specialized fields, and there are many of them out
there.
- Essays in Econometrics:
Collected Papers of Clive W. J. Granger (Econometric Society Monographs) (Volume 2): Clive Granger thought about many interesting time series topics, including
cointegration. This book includes a lot of his thoughts.
- Spectral
Analysis of Economic Time Series (Princeton Studies in Mathematical
Economics): Another more general book on time series by Clive
Granger. (It is also out of print.)
- Books about longitudinal/panel data
- Books about data mining
- Books about general statistics and econometrics
- Math books (that can be helpful for statistics and econometrics proofs)
- Matrix
Differential Calculus with Applications in Statistics and Econometrics,
2nd Edition: Professor Hurvich recommended this book when I realized
that I would have to compute standard errors based on the Hessian from
maximum likelihood optimization, based on a reparameterization.
- Matrix Analysis and
Topics in Matrix
Analysis (both by Horn and Johnson): I came across these books on a quest for more information about the eigenvalues of Hadamard products of matrices. These
books thoroughly cover advanced linear algebra - the theorems that are well-known to people in the field and that would be helpful for people using linear algebra to
prove theorems, but that are not generally taught in introductory linear algebra classes. They are very handy to have around.
- Regular
Variation (Encyclopedia of Mathematics and its Applications): This
is a classic book on regularly varying functions. These functions
(particularly slowly varying functions) are very handy for
semiparametric descriptions of long memory time series (since the "short
memory" part can be described by a slowly varying function instead of
parametrically).
- Computer science books
If you are looking for information on my research, please see my NYU webpage.
If you are looking for a tutor in any of these subjects and you are
located in the New York/New Jersey area, please contact me.
You can read more
about me on my webpage.
Comments?