STAT 391 Spring 2021

About the course

Information specific to the on-line format is in red.
Instructor: Marina Meila
Padelford B-321, Box 354322
543-8484 (in Padelford)
mmp@{cs.washington.edu,stat.washington.edu}
Office hours: Instructor: TBA,
TA:
TA: Zhaoqi Li
NetId: zli9
Lecture place and time: Tue, Thu 3:30 - 5:20 on-line

Course home page: http://www.stat.washington.edu/courses/391/spring21.

Class mailing list: stat391a_sp20
Textbooks: Extensive course notes are available for download. Other recommended books are listed on the books web page.

Format: The course will consist of two weekly 2 hour lectures and weekly homework assignments.

For the first time this year, occasionally, the 4th lecture hour on Thursday will be devoted to special topics, on which you will not be tested (except possibly for extra credit).

Lectures will be on Zoom. The links for lectures are posted on Canvas (go to Zoom in the left hand menu). Lectures will be recorded. Slides will be posted on the Handouts/Course notes page ahead of the class; during lectures I will use these slides as background, plus a virtual whiteboard, with GoodNotes. After class, the annotated slides will be posted on Canvas (go to Files/lecture notes).

Participation: Attendance is not mandatory. However, class participation is important and therefore it will be part of your grade. By class participation I mean more than just being in class; you're expected to actively participate, either by asking questions (the easiest) or by answering my questions, my invitations to comments, or questions on the discussion board.
During lectures: Ask questions in class (chat OK). Reply to the questions I pose during lectures. On the discussion board: Reply to questions asked by others on statistics or other course topics. Replies to non-homework related questions count more.
How much participation is enough? Once a week on average, either in class or on discussion board, is plenty.
Assignments: Typically assignments will be posted on the web (usually) on Thursdays and will be due before class (usually) the following Thursday. The homework should be submitted as a single pdf file. We do not grade homework answers that are in the code.

The homeworks will consist of problems and short programming assignments. It is important to turn in your homework timely. First, because homework is a component of learning the new material. New lectures will build on past lectures as well as past homework results. Second, fair grading requires fair conditions for everybody. You are allowed to be late with your homework twice with no penalty. Further late homework will be penalized by up to 50%. No late homeworks are accepted once the solutions are out or the graded homework returned (this happens typically one week after the homework due date).

Programming assignments:A typical programming assignment is: generate some data from a given distribution, do some processing, plot the results, draw conclusions. You will have to implement the data generation and processing. The programming language is your choice, and you are free (and encouraged!) to use library functions, but you have to write the code for the algorithms that are the subject of the course. For example, you are not expected to implement your own random number generator, but you will have to write code (that will use the random numbers) to simulate a dice roll. You will implement the formula for mean and standard deviation even if the software you are using has a built-in function for that.

Occasionally, you may be required to submit the code you used for the programming assignments. This should not be in the same file with the homework, but submitted separately. We will not grade or debug your code, but we will do random checks to see if the code indeed does what it's supposed to do.

Python and Matlab. We will offer some support for those of you who want to learn one of these languages while taking 391. The TA will offer a pyton tutorial, and a short intro to Matlab will be posted. Matlab is very easy to learn on your own, and is awesome at making plots. Python is very popular overall, and also very easy to learn. Making plots is comparatively less easy, but the package matplotlib (which emulates the Matlab plotting capabilities) is a great help. The virtual classroom environment we will make available will have both Matlab and python installed for you. We will have to confirm that the virtual classroom is available this quarter. Making plots. Often, the result of your programming assignment will be displayed as a plot. You are allowed to use any programming environment for making the plots. But, since in the past students experienced some difficulties with this, here are some suggestions. (1) Matlab is very easy to use for making plots, and has some nice functionality for special plots like histograms. (In addition, matlab is available on the CS clusters and on many other machines in CS -- not sure this is true any longer.) (2) python (matplotlib) is almost as nice as matlab (3) In past years, many students who didn't use Matlab used Excel. I noticed from their homeworks that, towards the end of the class, they were having more difficulty than the Matlab/python users in getting their plots to look good (and had more frustaration with the homework). (4) There are other programs around, like R, Splus, LaTeX, ...

Other homework problems will require calculus and easy arithmetic (in addition to what you learn in this class, of course). Both literal and numeric answers should be brought to their simplest form for full credit, and to allow you to fully appreciate the result. Math topics you should review: set theory, combinatorics and counting, taking derivatives and their meaning , integration (integration by parts, elementary properties of integration, primitives of common functions), elementary multivariate calculus and matrix operations. If there is need, tutorial sessions can be arranged for any of these topics.

Teamwork: Students are encouraged to talk to each other, to the TA, to the instructors, or to anyone else about any of the assignments. Any assistance, though, must be limited to discussion of the problem and sketching general approaches to a solution. Each student must write out his or her own solutions to the homework, including the code. We may use plagiarism detection software. v
Grading: The grade in STAT 391 this spring will be 5-10% participation, and the rest homework, possibly including a larger homework that we will call a project. I am considering having an optional oral exam. If the exam is given, and you choose to take it, the exam grade will replace/reweigh your homework part of the grade (details later). Update 6/5/2021: The exam will be given, but it will not be part of the grade.
Prerequisites: A class in multivariate calculus (partial derivatives, multiple integrals, matrix algebra). A class in probability, including conditional probability and independence. Ability to write simple programs in order to do the homework. Ability to reason mathematically, to read, understand and discover proofs. Here is a list of the calculus topics we typically use.

Marina Meila
Last modified: Thu March 22 2018