Statistics of Experimental Design

1
Statistics of Experimental Design
wi2144st
2
(No Transcript)
3
Aim

• Introduction probability and statistics
• Familiarise engineering students with the jargon
  of the statisticians
• Model-based analysis of experimental data
• Design of experiments

4
Contents
Probability
Example 1 Probability of success in test
Distributions
Descriptive Statistics
Estimation theory
Example 2 Probability of success in test 2
given that test 1lt5.5?
Hypothesis testing
Linear Model
Non-linear regression
Answer 0.55 and 0.15
Design of experiment
Model selection/discrimination
5
Contents
Probability
Distributions
Descriptive Statistics
Estimation theory
Hypothesis testing
Linear Model
Non-linear regression
Design of experiment
Model selection/discrimination
6
Contents
Probability
Distributions
Descriptive Statistics
Estimation theory
Hypothesis testing
Linear Model
Non-linear regression
Design of experiment
Model selection/discrimination
7
Contents
Descriptive Statistics
Probability
Example What is mu and sigma?
Distributions
Estimation theory
Estimation theory

• Bias
• Robustness
• Confidence Interval
• Bootstrap methods

Hypothesis testing
Linear Model
Non-linear regression
Design of experiment
Model selection/discrimination
8
Contents
Descriptive Statistics
Probability
Example 1 When you have less than 4. 5 on
test 1, you will not pass
Distributions
Estimation theory
Hypothesis testing
Example 2 Average Test1Average Test 2
Linear Model
Non-linear regression
Answer No and no
Design of experiment
Model selection/discrimination
9
Contents
Descriptive Statistics
Probability
Distributions
Estimation theory
Hypothesis testing
Linear Model
Non-linear regression
Design of experiment
Model selection/discrimination
10
Contents
Descriptive Statistics
Probability
Distributions
Estimation theory
Hypothesis testing
Linear Model
Non-linear regression
Design of experiment
Model selection/discrimination
11
Contents
Descriptive Statistics
Probability
Distributions
Estimation theory
To improve estimate ... To improve
prediction of model
Hypothesis testing
Linear Model
Non-linear regression
Design of experiment
Model selection/discrimination
12
Contents
Descriptive Statistics
Probability
Given Data Choose between models
Distributions
Estimation theory
Hypothesis testing
Given Different Models Choose between
experiments
Linear Model
Non-linear regression
Design of experiment
Model selection/discrimination
13
The Lectures

• Introduction, probability
• Probability distributions
• Simulation, expectation and variance
• Joint distributions
• Central limit theorem, data exploration
• Estimation
• Hypothesis testing
• Stochastic vectors linear model and estimation
• Linear model, confidence interval and hypothesis
  testing
• Linear model, Non-linear regression
• Non-linear regression
• Experimental design
• Model selection
• Review

14
Overview
Data Model Experimental Error
Entities
Actions
Characterisation Experimentation Estimation Hypoth
esis testing
to be continued
15
Course Material and Information

• Part I A Modern Introduction to Probability and
• Statistics by Dekking et al, ISBN 1-85233-896-2
• (Autumn 2004 and later versions of Kanstat
  lecture notes still usable.)
• Part II Lecture Notes, plus exercises
• Exercises and Examination Set Edition 7 available
  as PDF
• Data for exercises and past examinations
• Software
• SSORstat
• PastiFit
• PastiMos
• Excel
• . any other, eg Splus, SAS, SPSS, Matlab

16
Where, what
VSSD A Modern Introduction to Probability and
Statistics by Dekking et al, ISBN
1-85233-896-2
Computer rooms Data for exercises and past
examinations Software SSORstat, PastiFit,
PastiMos and Excel
Web page http//dutiosc.twi.tudelft.nl/a90
(also via BB page) Data for exercises and past
examinations (zip-file) Exercises and Examination
Set (1994-2006) as pdf Lecture notes Statistiek
van Proefopzetten as pdf Download software
SSORstat, PastiFit, PastiMos
And of course e-mail to the lecturers L.E.Meester
_at_tudelft.nl, P.J.T.Verheijen_at_tudelft.nl
17
Course Material and Information
Changes with respect to 2005/2006
Lecture notes Statistiek van proefopzetten ONLY
available as pdf on website. Exercises and
Examination Set minor updates and inclusion of
exams 2004/2005 ONLY available as pdf on
website.
18
Timetable

• Course schedule 3rd period Room B
• Thursday hour 12 Lectures (/Exercises)
• Friday hour 567 Lectures/(Computer) Exercises
• Course schedule 4th period Room B
• Wednesday hour 234 Lectures/(Computer)
  Exercises
• Thursday hour 34 Lectures/Exercises
• Examination schedule
• Full course examination Thu., June 14, 14-17
• Full course examination Tue., August 28, 14-17

19
Examination

• Material present
• Computer with
• SSORstat
• PastiFit
• PastiMos
• Excel
• Formulas (from appendices Lecture Notes)
• Allowed normal calculator

20
Examination
21
and what else
Dans le domaine de la science, le hasard ne
favorise que les esprits qui ont été
préparés. Louis Pasteur, 1822-1895