Computational Statistics

Spring semester 2018

General information

Lecturer	Marloes Maathuis
Assistants	Peter Hinz, Loris Michel, Claude Renaux
Lectures	Thu 13-15 HG F 3 >> Fri 09-10 HG G 3 >>
Exercises	Fri 10-12 HG F 3 >>
Course catalogue data	>>
Credit points	In order to obtain ECTS credit points, you need to pass the official exam. In order to obtain ETH credit points (this only applies to doctoral students from certain departments), you need to solve and submit at least 70% of the exercises.

Course content

We will study modern statistical methods for data analysis, including their algorithmic aspects and theoretical properties. Special emphasis will be placed on re-sampling based methods for inference. The course is hands-on, and methods are applied using the statistical programming language R.

The main text for the course is the following script. We will also use material from books (see Literature). The material for each week will be indicated under the Course material tab.

The exam will cover all material that is discussed during the lectures and the exercise sessions. If you miss a class, please make sure to copy class notes from someone else.

Announcements

First week
Beginning of lecture: Thursday, 22/02/2018.
Exercises start on 23/02/2018 at 10:15 in room HG F3, with a special introduction to the R software. Please bring a laptop. See the Exercises tab and further helpful links related to R.
Special dates:
- Different room: The class on March 29th takes place in HG E 3.
- Holidays: There will be no class on March 30th, April 5th, April 6th and May 10th.
- Cancellation: The class and exercises on May 11th are cancelled.

Course material

Week	Topic
Week 1	Linear regression Slides 1, R Code 1 Reading: Script Sections 1.1-1.6, ISLR Sections 3.1-3.3
Week 2	Linear regression Slides 2 (updated March 8), R Code 2 Reading: Script Sections 1.1-1.6, ISLR Sections 3.1-3.3
Week 3	Confidence intervals, and bias-variance trade-off Slides 3, R Code 3 Reading: ISLR 2.2.1-2.2.2, 3.5
Week 4	KNN, Cross validation Slides 4, R Code 4a, R Code 4b Reading: Script Ch 4, ISLR Ch 3.5+5.1
Week 5	Bootstrap R Code 5 (updated March 29) Reading: Shalizi Ch 6, script 5.1 - 5.2.4 + 5.4 Optional reading: Paper by Andrews, Paper by Hesterberg
Week 6	Bootstrap R Code 6 Reading: Shalizi Ch 6, script 5.1 - 5.2.4 + 5.4 Optional reading: Paper by Andrews, Paper by Hesterberg
Week 7	no class
Week 8	Bootstrap and Monte Carlo tests Slides 5, R Code 7a, R Code 7b, R Code 7c
Week 9	Permutation tests Slides 6, R Code 8
Week 10	Multiple testing and model selection Slides 7, Slides 8, R Code 9 Reading: ISLR Ch 6.1 and 6.2 Script: Section 7.7
Week 11	Inference after model selection, moving beyond linearity R Code 10, R Code 11
Week 12	no class
Week 13	Beyond linearity Slides 9, R Code 12 Reading: ISLR Ch 7 Script: Sections 3.3, 3.4 and 4.5
Week 14	Tree-based methods Slides 10, R Code 13 Reading: ISLR Ch 8
Week 15	Bagging and boosting Slides 11, R Code 14 Reading: ISLR Ch 8

Exercise classes

The exercises form a very important part of the course. They are also important for exam preparation, since part of the exam requires the use of R. If possible, please always bring a laptop to the exercise classes. After a brief introduction by one of the assistants, you can work on the exercises yourself or with others, and ask any questions you may have (about the exercises or the course material). There will be several assistants available to answer your questions.

Unless you are a PhD student who requires ETH credit points, you do not have to hand in your exercises. We will provide solutions, and you are expected to check your own work. Please ask if anything is unclear, for example if you don't understand the solutions, or if you found a different solution.

PhD students who require ETH credit points should place the solved exercises in the corresponding tray in HG J68 on the due date by noon at the latest.

Exercise sheets

The new exercise sheet will be uploaded on Tuesday preceding the corresponding session.

Exercises	Discussion	Deadline	Slides / Notes
	February 23, 2018		R-Tutorial, R Code
Series 1	March 2, 2018	March 9, 2018	R Code for discussion
Series 2 (updated March 19)	March 9, 2018	March 15, 2018	R Code for discussion
Series 3 (updated March 26)	March 16, 2018	March 22, 2018	R Code for discussion (Shalizi, Appendix N)
Series 4 (updated March 26)	March 23, 2018	March 29, 2018	R Code for discussion
Series 5, R-skeleton	April 13, 2018	April 19, 2018	R Code for discussion
Series 6	April 20, 2018	April 26, 2018	R Code for discussion
Series 7 (updated May 18), data	April 27, 2018	May 3, 2018	R Code for discussion
Series 8	May 4, 2018	May 17, 2018	R code (html) for discussion
Series 9	May 18, 2018	May 24, 2018	Subdifferentials
Series 10 (revised on 31.05.18)	May 25, 2018	May 31, 2018
General discussion	June 1, 2018

Solutions

The solutions will be sent weekly to the students enrolled for this course.

Literature

Main literature

P. Bühlmann, M. Mächler. Script Computational Statistics. (Version of October 12, 2016).
G. James, D. Witten, T. Hastie, R. Tibshirani. An Introduction to Statistical Learning: with Applications in R [ISLR]. Springer.
T. Hastie, R. Tibshirani, J. Friedman. The Elements of Statistical Learning [ESL]. Springer.

Other relevant literature

C. R. Shalizi. Advanced Data Analysis from an Elementary Point of View.
B. Efron, T. Hastie. Computer Age Statistical Inference. Cambridge University Press.
J. E. Gentle. Elements of Computational Statistics. Springer.
W. N. Venables, B. D. Ripley. Modern Applied Statistics with S. Springer.
L. Wasserman. All of statistics. Springer.