lu.se

# Institute/Department:

## COURSE CODE CHANGED TO FMSF65MASC05 the new course page can be found here.

Mathematical Statistics, Mathematics Centre

Credits:
7.5 credits (10 ECTS credits)

Prerequisites:
Second course in probability theory, corresponding to MASC01 (MAS203), is recommended.

Requirements for assessment:

Written exam, written project report, and compulsory computer labs

Time and place:

• The course will be given in English.
• Course start: Monday March 20, 8:15-10:00 in E:B
• The schedule for the course can be found here.

Course literature:

Main course litterature:

• "Statistics for Experimenters", by Box, Hunter and Hunter. Wiley, 2nd ed. (2005)

An alternative book is Montgomery "Design of experiments". We will however use only the main book by Box, Hunter and Hunter, for all lectures and exercises.

Extra course material:

Computer labs:

The computer labs are compulsory for the course. Labs will be done in the package R. You can download this for free and install on your computer. Can be found the R's project homepage.

Project:

An individual project with written project report shall be completed during the course.

Examination:

Written exam Monday May 29th, 08:00-13:00 in Sparta:B and Sparta:D

Example exam: Exam 2016 and Exam Assistance Sheet

## Lectures

There will be 7 lectures on this course. All chapter numberings refer to the course book by Box, Hunter and Hunter (2nd edition)

1. Chapter 1-3.2. Paper on yeast growth.
2. Ch 3. Extra material on comparison tests-confidence intervals (Anevski, 2012)
3. Ch 4. ANOVA, Graphical ANOVA.
4. Ch 4. Randomized blocks. Latin squares.
5. Ch 4. 5. Balanced incomplete designs. Factorial designs.
6. Ch 6.1-6.8 Fractional Factorial Design
7. Ch 10 Multivariate Regression

## Exercises

There will be seven exercise sessions on this course. All numberings refer to the book Box, Hunter and Hunter (2nd edition)

1. Problems for chapter 2: 1-6, 12. Problems for chapter 3:  1
2. Problems for chapter 3: 4, 5, 7, 8, 11, 22, 23, 26
3. Chapter 4: Exercises 4.1, 4.2. Problems 1
4. Chapter 4: Exercises 4.4 (without computer). 4.5, 4.7, 4.8. Problems 2.
5. Chapter 5: Problems 5.1-5.5.
6. Chapter 6: Problems 6.1, 5.8, 4.3

## Computer labs

There will be seven computer labs. Attendance and completion of all computer labs is compulsory for the course. Remember to sign up for each lab with Maria.

1. LAB 1.
2. LAB 2.
3. LAB 3. Datafiles to use: mollusc penicilin
4. LAB 4. Datafile to use: BHHp8ch5
5. LAB 5
6. LAB 6

## Project

1) Find a dataset within a topic that you are interested in or consider an experiment that would give you data about a phenomenon that you are interested in.

2) Study as much as you can about the dataset or phenomenon: its origin, potential sources of systematic error, potential sources of noise.

3) Formulate one or more hypothesis that can be investigated from the data.

4) Structure an experimental design for your phenomenon and carry out the experiment, or if your starting point is an existing dataset: structure an "experimental design" around the data suitable for your hypothesis.

5) Carry out relevant hypothesis testing.

6) Validate and discuss your assumptions.

8) Write a short report on 1) through 7) and send it to me (sva@maths.lth.se)

Do feel free to carry out the project either as individual or as a small group, subject to your own preference.

Once you have decided on a phenomenon and experiment, or on a dataset (or if you are simply contemplating one) do drop me an e-mail so I know where you are in the process. Also do feel free to contact me along the way for discussion on your data, hypothesis, choice of tests, validation, and results.

If you have not yet been in contact with me regarding your project, do send me an e-mail. Also if you are unsure what dataset to start from - in worst case I will suggest you an experiment to start from.

Please do aim to send me your final project report before June 3rd. 2017

Lectures:

Søren Vang Andersen,

email: sva@maths.lth.se,

tel: +46 46 222 47 47

Room: MH:244

Labs and Exercises: