.
.
S-87-75
SIMON ?
FRASER
UNIVERSITY
MEMORANDUM
TO: ?
Senate ?
FROM:
?
J.W.G. Ivany,
Chair, SCAP
SUBJECT: Graduate Curriculum
?
DATE: ?
Nov. 19, 1987
Changes - Math & Statistics
Reference: SCAP 87-46
Action undertaken by the Senate Committee on Academic Planning/Senate
Graduate Studies Committee gives rise to the following motion:
MOTION: ?
"That Senate approve and recommend approval to
the Board of Governors, as set forth in S.87-75
the following new course:
STAT 602
?
Generalized Linear and Non-linear
Modeling"
U^
S11101 FRASER t1'E'.1T?
New Graduate CnsrRc
Prirnoal vorm
CALENDAR INFORMATION:
Department: ?
Mathematics and Statistics ?
_Course Number:
Stat602
Title:
?
Generalized Linear and Nonlinear Modelina
A methods-oriented
unified approach to a broad array of non-linear regression
Description:
modellin
g
methods
including c1aical rPgr
cirn
r
1nc j if4r .
raqxession,
probit analysis, dilution assay, frequency count analysis, ordinal-type responses,
and survival data. Aoroject will beigned
relat p
a
tg,=4Wts' field of
Credit Hours:
?
-
?
Vector:
?
602-3
?
_Prerequiaite(c) if env:
Stat302
or Stat 330 or permission of instructor. Open only to graduate students in departments
other than Mathfliatics and Sft3
1-i
- ?
-
rNRou.F.NT ANI) SCIIEDULINr.:
Estimated Enrollment:
?
6
?
When will the course fIrst be offered: 1988 - III.
How often will the course be offered:
?
Once per year.
JUSTIFICATION:
Double list4ng with Stat 402. Available only to Graduate StudentJji
Departments other than Mathematics and Statistics.
RRSOURCES:
Which Faculty member will normally teach the courne:
Eaves,
Lockhart,_Routledge,
Stephens,
Swartz, Weldon
What are the budgetary implications of mountin
g
the course:
?
None -
Arc there sufficient Library resources (A
pp
end details):
?
Yes
Appended.
n)
Outline of the Course
(Course Description for Stat 402)
b)
An Indication of
the
comnetenco of the Faculty member to give the course.
c)
Library resources
Reserve copies of reference materials (see
course description Stat 402).
TiD..
Approved: Dcpnrtmefltfll Graduate Studies Committee:
Faculty Graduate uL48CO341ttPC
Faculty:
S.
?
Senate Graduate Studies
Commtttcc
Date:
tV
Pate: ?
87
Date:
not t
c:
Senate:
STATISTICS j902
?
Generalized Linear and Nonlinear Modelling
NOTE: This course extends the concepts, methods and
approach of STAT 302-3 to cover a wide variety of common
types of outcome data. It employs a modern unified appraoch
to a broad array of nonlinear regression problems.
1.
Brief review of fundamental background.
2.
Overview: Empty model, link function, simple examples
of structuring a mean value vector with link function
and design matrix, and of structuring variance with a
variance function; iterated reweighted least squares
estimation.
3.
Examples from exponential - type likelihood models:
Normal, including classical linear regression and other
links; Poisson, including log-linear regression;
Binomial, including lcsgit, prc
'
bit, and dilution assay.
Examples allowing overdispersic'n.
4.
Other examples.
5.
Inference: The variance-covariance matrix of the
estimated regression vector and confidence intervals for
linear predictors, fitted values, other
relevant
estimated quantities; comparative evaluation of models,
deviance, Pearson statistic, residuals.
6.
Logistic Regression.
7
?
Contingency tables and log-linear models.
8.
Ordinal-type outcome: Proportional odds model,
proportional hazards model.
9.
Experimental design: Randomized block, factori al
designs, latin squares.
10.
In addition to the STAT 402 outline, STAT 406 students will be expected
to analyse a data set from work in their own area of application.
Primar y
References
An Introduction to Statistical Modelling
Dobson
Chapman ?
Hall, 1983.
Other References: Sections 1 through 6 of:
The GLIM System, Release 3, Manual
Baker & Nelder
Numerical Algorithms Group, 1978.
Nonlinear Regression Modeling - A Unified Practical Approach
Rotkowsky ?
Marcel Dekker, 1983
.
.
I
SIMON FRASER UNIVERSITY ?
MEMORANDUM
To: Dr. John Webster, Dean of Science
?
From: Dr. George Bojadzicv,
FSGSC
Subject: STAT 602 Course proposal
?
Date: 2 October 1987
The following course proposal for STATistics 602 has been approved by the
Faculty of Science Graduate Studies Committee. Please include this item in the
agenda for the next Faculty of Science meeting.
ion
(available only to graduate students
in departments other than Mathematics & Statistics
Many graduate students engaged in quantitative research need the practice and mechanical
skills for analysing a wide variuety of common types of data which are beyond traditional
• classical regression and anova packages. STAT 602 will provide this. While these skills are
considered to be at the advanced undergraduate/beginning graduate level within the
discipline of Statistics, they should properly be credited at the graduate level for workers in
applications fields. It will require that the student already have some experience and
practical general sense of data frequency distributions, of experimental design, and of
classical regression/anova. However no further mathematical background will be needed.
The lecture components of STAT 602 and STAT 402 will be the same. Registration in
STAT 602 will be restricted to graduate students in departments other than Mathematics &
Statistics. They will be required to complete a substantial project analysing a data set in
their own field of interest.
0
