Dept. of Mathematics
Math 158 – Linear
Statistical Models
Spring 2008 (Second
Half)
Prof. Susan Martonosi
ANNOUNCEMENTS
* Add yourself to the course
email list. Send an e-mail message to listkeeper@hmc.edu. In
the body of the e-mail message, write subscribe math-158-l
* Please fill out the Student Information Form.
* Here’s a sample Project Report
LECTURE 2 FILES
LECTURE 4 FILES
Lecture 4 PowerPoint
LECTURE 6 FilES
Lecture 6 PowerPoint
LECTURE 8 FILES
Lecture 8 PowerPoint
LECTURE 10 FILES
Lecture 10 PowerPoint
LECTURE 12 FILES
Lecture 12 PowerPoint
LECTURE 14 FILES
Lecture 14 PowerPoint
HOMEWORK
ASSIGNMENTS Assigned
daily; due weekly at 5pm on Wednesday
DATA
SETS FOR KUTNER, NACHTSHEIM, NETER AND LI (.txt format; on-campus access only)
TEXT WEBSITE: http://apps.csom.umn.edu/Nachtsheim/5th/
PROJECT DETAILS – Note: If your project involves time series data, please read Chapter 12 of Kutner, Nachtsheim, Neter
and Li. We will likely not have time to
cover this chapter.
PROF.
MARTONOSI’S CONTACT INFO
Office: Olin 1277
E-mail: martonosi @ math . hmc . edu [please note the
spelling…]
Phone: ext. 7-0481
Office Hours: TBD
LECTURE
INFO
MW 1:15-2:30
Mondays: Beckman B134
Wednesdays: Parsons PC
Computer Lab
COURSE
GRADER
Tim Sweda
COURSE OVERVIEW (click here for day-by-day)
Linear regression models
form the foundation of statistical analysis, yet they are often misused and misinterpreted. In this
course, we will expand on the simple linear regression model presented in Math
62, and we will delve deeper into the theory and application of such models,
examining regression with several variables, polynomial regression, ANOVA,
model diagnostics and transformations, and variable selection. In addition to studying the theory of
regression, we will be analyzing real data!
Prerequisite: Math 62. 2 credit
hours.
AIMS
This course intends to
develop your understanding of the theory and application of statistical methods
through the study of linear regression models in particular. This is a first upper-level course in
statistics to present the foundations for statistical theory on which other
methods build. Your understanding of
linear models (and more generally, statistics) will be strengthened in two
directions: 1. Thorough theoretical
understanding of statistical models, assumptions and inferences; 2. Practical
understanding of the challenges of applying such models to real data. Indeed, at the heart of this course is the
interface between the statistical and probabilistic theory and its implications
on the practical use of statistics.
OBJECTIVES
On completion of this
course, you should be able to (among other things) construct an appropriate
regression model for a given set of data, including
a) identifying a model type
b) selecting explanatory variables
c) transforming the data
d) fitting the model
e) evaluating the applicability of the model to the data
f)
interpreting the
model
or identify when no such
model is appropriate.
TEXT
Applied Linear Statistical Models, 5th edition, by Kutner, Nachtsheim,
Neter and Li
ISBN 0-07-238688-6
Text materials website: http://apps.csom.umn.edu/Nachtsheim/5th/
Text
Data Sets (.txt format;
on-campus access only)
GRADING
Your grade in this course will be based on:
Homework assignments: 40%
Grading rubric for homework assignments:
Each problem will be graded on a 5-point scale:
1
point: Problem was attempted.
1 point: Student approached the problem correctly, irrespective of
technical mistakes that may have been made.
1 point: Any assumptions made were stated (if this criterion is not
applicable, then this point is a freebie for the student).
1 point: The solution is correct.
1 point: The writing and explanation are clear and easy to follow.
Under this rubric, it should be easy to get three
points out of five (for attempting the problem, stating assumptions and writing
clear explanations). Moreover, a correct solution might earn only three
points out of five if assumptions are not stated clearly or if the writing and
explanation are unclear. So I feel this scheme puts weight on the things
I'd like most to encourage.
The
homework in this course has two purposes: 1. to provide practice in the framing
and solving of practical problems; and 2. to develop your theoretical
understanding of the material. Thus your weekly assignments will consist
of both data analysis tasks and proofs.
The Homework
page will list daily the reading and homework problems corresponding to
that day’s lecture, but the homework will be collected weekly, at 5pm on
Wednesdays.
The best way to succeed in this course is to stay
current by
reviewing your class notes, reading the textbook and completing the homework
problems after each lecture. For
other suggestions for success, please read these Study Tips.
-
Homework
will be due at 5pm on Wednesdays.
-
No
late homework will be accepted without a note from a Dean.
-
Homework
must adhere to the Mathematics
Department Homework Guidelines format described at http://math.hmc.edu/~orrison/teaching/homework/
-
Homework
solutions will be posted on the course website.
-
You must work
individually on homework assignments, though you are encouraged to discuss with
your other classmates, provided you acknowledge any assistance given or
received.
-
Statistical
Software: Where indicated, you may
use statistical software of your choice to complete computational
problems. The PC and LAC labs on campus
have Minitab and SPSS installed on them.
Also available is a freeware known as R, which you can download from http://www.r-project.org/ Their website also provides documentation,
but a nice manual is by Peter Dalgaard: Introductory
Statistics with R, which is available at the library. Campus machines also have Maple and Minitab
installed, which both have statistical packages. Certain department computer labs may also
provide access to SPSS, SAS or Mathematica.
-
Extra Credit: Write a computer tutorial for an “alternative”
(i.e. non-Minitab) statistical software package – R, Matlab, Maple, SPSS, SAS,
or Mathematica (if you have access to these).
The tutorial should describe how to use that package to do a particular
statistical analysis that we have seen in class. In addition to the tutorial, you must write a
short summary describing the
strengths and weaknesses of the package you chose for doing that type of
analysis. If you elect to do any of
these extra credit tutorials, you are agreeing to allow me to use these
tutorials in future HMC courses; as such, you must also send me electronic
copies of the tutorials that I may edit.
Project:
30%:
In this
project, you will perform the entire regression model-fitting process by
finding an appropriate data set (either in an online database, from an
experiment you have already conducted or by designing your own experiment) and
conducting an appropriate regression analysis on this data using concepts
learned in the course. You will work in triplets
on this project, and all team members are expected to contribute equally.
Deadlines:
3/26
(in computer lab): Brainstorm project ideas with peers
3/31:
Written proposal due
4/16
(in class): Progress discussion with peers (each team member must bring a
preliminary model to the discussion and be prepared to explain it to others)
4/23:
Report draft due (each team member should bring a copy of the draft)
4/28
(in class): Discussion of drafts
4/30:
Final report due
Final Exam: 30%
The final will be an “in-class” exam during finals
week.
The homework is intended to
strengthen your theoretical understanding of the material. The project will help you to apply the theory
to a practical situation. The exam will
test your understanding of both contexts.
Note: you must receive credit on at least half of the homework assignments in order to pass the course.
If you need
disability-related accommodations, please discuss this with Dean Noda as soon
as possible.
STUDENT FEEDBACK
In addition to formal mid-course and end-of-course
evaluations, I encourage you to let me know immediately
after class if a concept was not clear or if there is something I can do to facilitate
the learning process in class. And if
you are struggling with the course, the best way to let me know is to come to
my office… frequently. I am here to help
you learn the material, but I cannot do so if you do not actively seek help.