San
Francisco Chapter of the American Statistical Association
Chapter
Elections Meeting
And
Student
Travel Awards
Date: June 2nd
Time: 4 p.m. – 6 p.m.
Place: VBT 219, California State University East
Bay, Hayward, CA
Chapter Elections: 4:00 – 4:30 p.m.
Nominees:
President Elect - Kit Lau
VP- Biostatistics - Ruixiao Lu
VP- General Applications - Clinton Brownley
Secretary - Jacqueline Shaffer
Treasurer - Doris Shu
Nominations are still open.
Refreshments: 4:30 – 5:00 p.m.
Student Presentations: 5:00 – 6:00
p.m.
Each
year the SF chapter of ASA gives student travel awards to support their
presentations at the following JSM.
This year each our recipients are:
Aida Yazdanparast and Tony Tran
Department
of Statistics and Biostatistics
California
State University East Bay
Effects of Oil Spills on Birds
Abstract
The Deepwater Horizon oil spill was an ecologically devastating
event in the Gulf of Mexico, which saw an estimated release of over 4 million
barrels of oil after flowing for three months in 2010. The impact of the spill
persisted despite the wellÕs capping. Understanding
the progression and making use of the details in this event is the first step
in planning to improve response time and reaction strategy for future disasters
relating to a local avian population. The aim of this project is to dynamically
illustrate the important features of the data set utilizing a blend of
analytics and graphics executed through R and Tableau software. In our data
set, we focus on 7,229 birds that were documented between May and October. We
will explore predicted probabilities as well as drill into the layers on a
variety of canvases to uncover relevant information unseen through raw data for
policymakers, environmental planners and ecology researchers, and others
fascinated by the importance of birds in our ecosystem.
Luca Pozzi
Division
of Biostatistics
University
of California, Berkeley
A Bayesian Adaptive Dose Selection Procedure
with a Count Endpoint
Abstract
In clinical drug development, a sequence
of studies is carried out to identify an efficacious and safe dose of a newly
developed pharmaceutical drug. Adaptive trial designs can considerably improve
upon traditional designs, by modifying design aspects of the ongoing trial,
including early stopping, adding or dropping doses, or changing the sample
size. In the present work we propose a two-stage Bayesian adaptive design for a
Phase II study aimed at selecting the lowest effective dose for Phase III. In
the first stage patients are randomized to placebo, maximal tolerated dose, and
one or more additional doses within the dose range. Based on an interim analysis, the study
is either stopped for futility or success, or enters the second stage, where
newly recruited patients are allocated to placebo, maximal tolerated dose, and
one additional dose chosen based on interim data. Assuming a monotone
dose-response relationship, at interim, criteria based on the Predictive
Probability of Success are used to decide on whether to stop or to continue the
trial, and, in the latter case, which dose to select for the second stage. In
addition, at interim, criteria based on the Predictive Probability of Success
are used to decide on whether to stop or to continue the trial, and, in the
latter case, which dose to select for the second stage. Finally a dose will be
selected as lowest effective dose for Phase III either at the end of the first
or at the end of the second stage.
Rahul Mazumder
Department
of Statistics
Stanford
University
Regularization methods for learning large
incomplete matrices
Abstract
Abstract: In many real
life applications, available data is in the form of a large matrix (say users
and items or genes and patients, etc) with many
entries missing. This is popularly dubbed as the matrix completion
problem. The task is to come up with predictions for the missing entries
under certain meaningful assumptions on the underlying population matrix.
I will talk about models, efficient algorithms and applications for a certain
class of matrix completion problems --- for example convex relaxations of
low-rank models and variants accounting for robustness and data
uncertainty. If time permits I will also talk
about incorporating item-item similarities via graphical models within a
generic matrix factorization frame-work.
This is joint work with
my advisor Trevor Hastie (Stanford) , Robert Tibshirani (Stanford) and Deepak Agarwal
(Yahoo Research !).
Directions and campus Map can be found at:
http://www20.csueastbay.edu/about/visitor-information/maps-campus-locations/hayward-campus-map/
The VBT building is closest to parking lots F and G. Parking is $2/hour ($10/day) and is enforced 24/7. Details about the parking is at the bottom of the campus map.
Return
to Bay Area ASA Homepage