MIME-Version: 1.0 Content-Type: multipart/related; boundary="----=_NextPart_01D155C7.922E8470" This document is a Single File Web Page, also known as a Web Archive file. If you are seeing this message, your browser or editor doesn't support Web Archive files. Please download a browser that supports Web Archive, such as Windows® Internet Explorer®. ------=_NextPart_01D155C7.922E8470 Content-Location: file:///C:/D199C656/index2016.htm Content-Transfer-Encoding: quoted-printable Content-Type: text/html; charset="windows-1252"
Statistics 248 - Some tentative Details - Spring<=
/span>
2016
Analysis of Time Ser=
ies
Subtitle. Random Process Data Analysis: Models =
and
Methods
Meant for. Graduate students in Statistics and oth=
er
Departments.
Instructor. David Brillinger, brill@stat.berk=
eley.edu
Office hours: Wednesday 15:00 - 18:00 (or just before sunset if it is earli=
er)
in 417 Evans
GSI. Suzette Puente
Office Hours TBA
Classes. Tu Th 11:00 - 12:30 in 330 Evans
Lab
Section. Friday 15:00 to 17:00 334 Evans [May change=
]
Lab
website. TBA
Course
Homepage. www.stat.berkeley.edu/~brill/Stat248/index2=
016.html
See also
248 bCourses site
The grade. Will come from a combin= ation of a scientific paper proposal and an independent scientific paper.<= o:p>
The proposal is due Tuesday March
29 at class and the paper is due Monday May 9, by 3pm, under my office door, 417 Evans or in 367 Evans.=
Prerequisites. Some probability and statistical infer=
ence
background
The course.=
b> Time series, {X(t)}, will be defined in a broad manner. In particular “time” may
take values in Rk or be function-valued. The values=
of X may be in =
other
than Rl
Course materials will be directed towards students carrying ou=
t an
analysis of pertinent scientific data using methods covered in the course (=
or
others agreed upon with the Instructor). The resulting scientific paper
constitutes the final exam for the course.
The lab.<=
/span> Wi=
ll
cover the real-valued time case, and material from the statistical pac=
kage
R, particularly the time series functions. It is meant to will the students
realize their time series papers.
R
time series issues. See Stoffer’s www.stat.pi=
tt.edu/stoffer/tsa2/Rissuess.htm
Sylabus. [In preparation]
Some pertinent books=
:
R.
H. Shumway and D. S. Stoffer (2011). Time Series Analysis and Its Applications with R Examples. Available electronically via OskiCal and
by purchase from Springer for $24.95.
Bloomfield=
, P.
(2000). Fourier Analysis
of Time Series: an Introduction, Second Edition. Wiley
http://onlinelibrary.wiley.com/book/10.1002/0471722235
D. R. Brillinger,
Time Series: Data Analysis and Theo=
ry,
SIAM. Particularly Addendum
Available electronically vi=
a OskiCal
Cryer,
J. D., Chan, K-S.
(2008). Time Se=
ries
Analysis with Applications in R, Second Edition. Springer.
Douc, Moulines and Stoffer (201=
4) Nonlinear Time Series
Dutileul, P. (2011=
). =
Spatio-Temporal Heterogenei=
ty.
Cambridge
Gelfand,A. E., Diggle, P. J., Fuentes, M., Guttorp, P. (2010). Handbook of S=
patial
Statistics. CRC.
P. Guttorp, Stochastic Modelling of Scientific Data, Chapman and Hall.
Available electronically: http://www.download-genius.com/download-k:Stochastic%20M=
odeling%20of%20Scientific%20Data.html?aff.id=3D6253 [In preparation]
Shumway and Stoffer EZ
Venables, W. N. and Ripley,B. D. (2003). Modern
Applied Statistics with S
http://download.springer.com/static/pd=
f/619/bok%253A978-0-387-21706-2.pdf?auth66=3D1391281342_9752ec6ca873eb9e0e7=
7056543b3cd87&ext=3D.pdf
Proposal
and paper details.
1. Due on March 29, =
or
earlier, is a 1-2 page proposal setting down -
a) a
brief description of the data set you intend to analys=
e.,
b) an
indication of the source of the data set,
c) the
objectives of your investigation,
d) the
analyses you anticipate completing.
These are to be brie=
f.
The point is that the instructor can interact with you a bit before you do a
lot of work.
2. Due on May 9, or
earlier, under my office door, 417 Evans, is your paper.
Please submit as a o=
ne-sided
hard copy.
3. The course grade =
will
come from your paper and proposal.
4. Exam. Your paper =
will
constitute the course final.
The Paper
0. Have a title,
sections and a summary.
START YOUR PAPER WITH
"THE question that is will be considering in this =
study is ..."
END YOUR PAPER WITH "My answer to the
question is ..."
1. Please hand in a =
hard
copy. Have it =
≤12
pages, double spaced, point size &=
#8805;12pt,
single column, one-sided and one inch or larger margins.
2. As indicated abov=
e, start
out your paper with the scientific question you will be addressing.
3. Describe the
important parts of your analyses. Unless the instructor agrees to something
else, use only methods discussed in class or in Cryer & Chan or in Gelfand et al or in Guttorp or in
Shumway & Stoffer or in the DRB book.
4. Be
specific, clear, factual.
5. Indicate details =
and
sources of your data.
6. Provide the answe=
r to
your question and implications:
i) with subject-ma=
tter
interpretation as possible,
ii) that
are properly qualified (for the sceptical reade=
r -
here, me)
7. Lay
out any final models (with uncertainties for parameter estimates as possibl=
e)
8. Mention especially
important points (eg. model limitations, unexpe=
cted
results, and suggestions for future studies.)
9. Include important=
computer
commands in an Appendix. (These will be examined if there is a need and will
not count in 12 page limit.)
10. Include a list of
references.
11. If you analyze an
ordinary time series, carry out both time- and frequency-side analyses.
12. For your paper
provide some comparative discussion of the analyses, e.g. the time-side and=
the
frequency-side results.
Some specific sugges=
tions
re the paper
1. Be
clear what the question you are addressing is. Remember to provide a clear
answer.
2. Check the basic
assumptions (eg. stationarity, no outliers pres=
ent)
by plotting the data, getting stem-and-leafs, etc.
3. Think about the
available EDA methods, eg. =
re-expression
of the origibal data.
Some previous papers.
Characteristics of variable frame ra=
te
videos
Potential fields in predicting prima=
te
reaching behavior
Does cell-specific synapse strengthe=
ning
induce cell-specific changes in neural dynamics in the
corticostriatal circuit?<= o:p>
A preliminary time series analysis of
bitcoin prices
Time series analysis of trader
inventories
How bad is distracted driving?
What is missing in the Newton-Euler
dynamics: Time series analysis of mismatched sensing in robot systems=
span>
Time and frequency domain analysis of
climate impacts on evapotranspiration
Exploration of a relationship between
electricity production from fossil fuels and renewable sources using time
series analysis
Modeling vehicle thefts in Oakland
Can simulated earthquakes capture the
key features of engineering interest found in recorded earthquakes?
Why, is automatic speech recognition=
far
worse than human? Exploratory data analysis of speech signal and diagnosis =
of
acoustic model
Comparison on slip rate pulsing from
repeating earthquakes to geodetic data in creeping section of the San Andras Fault
A high frequency time series analysi=
s of
GOOG stock volatility during flash crash 2010
Time-series analysis for traffic flo=
w of
Cory Hall elevators
Epidemic type aftershock sequence mo=
del
describing the Gulf of California and southern San Andreas fault
zone
Computational fluid waves analysis a=
nd
simulations
Failure identification in single
overhead cam
Can the seasonality in Tibetan
seismicity be explained by precipitation?
A time series analysis of the traffic
flow on the Bay Bridge
Time series analysis of human joint
movement
Time series analysis of chiller
operations
Pitch detection
NOTE. Materials will be added to this page, and some changes <=
span
class=3DGramE>introduced as=
the
semester progresses.
23 January 2016