|
A CSCAR Workshop
Center for Statistical Consultation and Research,
The University of Michigan
October 10, 2008
Traditional regression analysis for continuous data fits a single linear
function of the independent variables through the center of the observed
response, under the normal distributional assumption. In practical applications,
one often encounters situations that require substantial modification of the
standard methodology. This one-day workshop discusses certain advancements in
the area of regression that goes beyond the realm of the traditional framework.
Specifically we shall cover topics such as
lowess, spline, and
quantile regression. Spline
and lowess techniques involve
fitting piecewise models in an attempt to
identify the relationship between the
outcome and predictors under a robust,
distribution-free framework. On the other
hand, quantile regression is a model of the
linear association between predictors and
quantiles of the response. This technique
is particularly useful for heterogeneous
populations where interest lies on a
percentile rather than the center.
In
addition, we shall also discuss regression
procedures for zero-inflated models that
are applicable to count data with an excess of
zero counts. Methodological modules will be
interspersed with hands-on computer
implementation using the statistical software
package Stata (Version SE 10.0 for Windows).
Instructors:
Ananda Sen, Associate Research Scientist at CSCAR.
and Adjunct Associate Professor at the Department of Statistics.
Joe Kazemi, Statistical Consultant
at CSCAR.
Prerequisite:
A Windows instruction class or equivalent
knowledge, basic experience in data
analysis and regression, and a basic
knowledge of Stata.
Provisions:
The enrollee will receive lecture notes. Morning refreshments will be served. Break time for lunch (lunch not provided).
When:
Friday, October 10,
2008
9:00 A.M. -
5:00 P.M.
Location:
Rackham
Bldg, 2nd floor, North Alcove in
the West Study Hall
Size:
Maximum of
15 participants.
Fee:
Registration until September 26, 2008:
$150 for University of Michigan affiliated
faculty, staff and students
$325 for others
Registration after September 26, 2008:
$180 for University of Michigan affiliated
faculty, staff and students
$390 for others
Please make check payable to CSCAR-University of Michigan, or give the University of Michigan Project/Grant or
short code to be billed. Send check to CSCAR, 3550 Rackham Bldg., University of Michigan, 915 E. Washington St., Ann Arbor MI 48109-1070.
Registration:
Call CSCAR (734) 764-7828. Enrollment is limited.
|
