Upload
duongdan
View
219
Download
1
Embed Size (px)
Citation preview
1
9/26/2004 Stat 567: Unit 14 - Ramón León 1
Unit 14: Introduction to the Use of Bayesian Methods for Reliability
Data
Notes largely based on “Statistical Methods for Reliability Data”
by W.Q. Meeker and L. A. Escobar, Wiley, 1998 and on their class notes.
Ramón V. León
9/26/2004 Stat 567: Unit 14 - Ramón León 2
Unit 14 Objectives• Describe the use of Bayesian statistical methods to
combine prior information with data to make inferences• Explain the relationship between Bayesian methods and
likelihood methods used in earlier chapters• Discuss sources of prior information• Describe useful computing methods for Bayesian
methods• Illustrate Bayesian methods for estimating reliability• Illustrate Bayesian methods for prediction• Compare Bayesian and likelihood methods under
different assumptions about prior information• Explain the dangers of using wishful thinking or
expectations as prior information
2
9/26/2004 Stat 567: Unit 14 - Ramón León 3
Introduction• Bayes methods augment likelihood with prior
information• A probability distribution is used to describe our
prior beliefs about a parameter or set of parameters
• Sources of prior information– Subjective Bayes: prior information subjective– Empirical Bayes: prior information from past data
• Bayesian methods are closely related to likelihood methods
9/26/2004 Stat 567: Unit 14 - Ramón León 4
Bayes Method for Inference
3
9/26/2004 Stat 567: Unit 14 - Ramón León 5
Bayes Theorem( ) ( )
( ) ( ) ( ) ( )
( ) ( ) ( )
( ) ( )
( ) ( ) ( )( ) ( )
1
1 2
|( | )
| |
||
|
where , ,..., is a partition of the sample space|
||
i ii n
i ii
n
P B A P AP A B
P B A P A P B A P A
P B A P AP A B
P B A P A
A A Af x f
f xf x f d
θ θθ
θ θ θ
=
=+
=
=
∑
∫
9/26/2004 Stat 567: Unit 14 - Ramón León 6
Updating Prior Information Using Bayes Theorem
4
9/26/2004 Stat 567: Unit 14 - Ramón León 7
Some Comments on Posterior Distributions
9/26/2004 Stat 567: Unit 14 - Ramón León 8
Differences Between Bayesian and Frequentist Inference
• Nuisance parameters– Bayes method use marginals– Large-sample likelihood theory suggest maximization
• There are not important differences in large samples
• Interpretation– Bayes methods justified in terms of probabilities– Frequentist methods justified on repeated sampling
and asymptotic theory
5
9/26/2004 Stat 567: Unit 14 - Ramón León 9
Sources of Prior Information
• Informative– Past data– Expert knowledge
• Non-informative (or approximately non-informative)– Uniform over range of parameter values (or
function of parameter)– Other vague or diffuse priors
9/26/2004 Stat 567: Unit 14 - Ramón León 10
Proper Prior Distributions
6
9/26/2004 Stat 567: Unit 14 - Ramón León 11
Improper Prior Distributions
9/26/2004 Stat 567: Unit 14 - Ramón León 12
Effect of Using Vague (or Diffuse) Prior Distributions
7
9/26/2004 Stat 567: Unit 14 - Ramón León 13
Eliciting or Specifying a Prior Distribution
• The elicitation of a meaningful joint prior distribution for vector parameters may be difficult– The marginals may not completely determine the joint
distribution– Difficult to express/elicit dependences among
parameters through a joint distribution.– The standard parameterization may not have practical
meaning• General approach: choose an appropriate
parameterization in which the priors for the parameters are approximately independent
9/26/2004 Stat 567: Unit 14 - Ramón León 14
Expert Opinion and Eliciting Prior Information
• Identify parameters that, from past experience (or data), can be specified approximately independently (e.g. for high reliability applications a small quantile and the Weibull shape parameter).
• Determine for which parameters there is useful informative prior information
• For parameters for which there is no useful information, determine the form and range of the vague prior (e.g., uniform over a wide interval)
• For parameters for which there is useful informative prior information, specify the form and range of the distribution (e.g., lognormal with 99.7% content between two specified points)
8
9/26/2004 Stat 567: Unit 14 - Ramón León 15
Example of Eliciting Prior Information: Bearing-Cage Time to Fracture Distribution
9/26/2004 Stat 567: Unit 14 - Ramón León 16
Example of Eliciting Prior Information: Bearing-Cage Fracture Field Data (Continued)
9
9/26/2004 Stat 567: Unit 14 - Ramón León 17
Example of Eliciting Prior Information: Bearing-Cage Fracture Field Data (Continued)
9/26/2004 Stat 567: Unit 14 - Ramón León 18
10
9/26/2004 Stat 567: Unit 14 - Ramón León 19
9/26/2004 Stat 567: Unit 14 - Ramón León 20
Joint Lognormal-Uniform Prior Distributions
( ) (log log )P X P Xσ σ≤ = ≤
12
9/26/2004 Stat 567: Unit 14 - Ramón León 23
Methods to Compute the Posterior
9/26/2004 Stat 567: Unit 14 - Ramón León 24
Computing the Posterior Using Simulation
13
9/26/2004 Stat 567: Unit 14 - Ramón León 25
Proof of Simulation Algorithm
( ) ( )( )
( )( )
( )
( )
0
0
0
0
00
0
P , accepted| accepted =
accepted
( ) ( )P , ( )
( )( ) ( )
( ) ( )( ) ( )
( ) ( ) ( ) ( )
( | )
PP
P U R f dU R
P U RP U R f d
R f dR f d
R f d R f d
f x d
θ
θ
θ
θ
θ θ θθ θ θ
θ
θ θ θθ θ θ
θθ θ θ
θ θ θθ θ θ
θ θ θ θ θ θ
θ θ
−∞∞
−∞
−∞∞ ∞
−∞
−∞ −∞
−∞
≤≤
≤≤ ≤
= =≤
≤
= =
=
∫
∫
∫∫
∫ ∫
∫
9/26/2004 Stat 567: Unit 14 - Ramón León 26
14
9/26/2004 Stat 567: Unit 14 - Ramón León 27
9/26/2004 Stat 567: Unit 14 - Ramón León 28
Sampling from the Prior
1 1log log logpt a U b= +
16
9/26/2004 Stat 567: Unit 14 - Ramón León 31
9/26/2004 Stat 567: Unit 14 - Ramón León 32
Comment on Computing Posterior Using Resampling
17
9/26/2004 Stat 567: Unit 14 - Ramón León 33
Posterior and Marginal Posterior Distributions for the Model Parameters
9/26/2004 Stat 567: Unit 14 - Ramón León 34
Posterior and Marginal Posterior Distributions for the Functions of Model Parameters
18
9/26/2004 Stat 567: Unit 14 - Ramón León 35
9/26/2004 Stat 567: Unit 14 - Ramón León 36
Simulated Marginal Posterior Distribution for F(2000) and F(5000)
19
9/26/2004 Stat 567: Unit 14 - Ramón León 37
Bayes Point Estimation
( ) ( )( ) ( )2g g f DATAθ θ θ−∫
9/26/2004 Stat 567: Unit 14 - Ramón León 38
One-Sided Bayes Confidence Bounds
20
9/26/2004 Stat 567: Unit 14 - Ramón León 39
Two-Sided Bayes Confidence Intervals
bounds
9/26/2004 Stat 567: Unit 14 - Ramón León 40
Bayesian Joint Confidence Regions
21
9/26/2004 Stat 567: Unit 14 - Ramón León 41
Bayes Versus Likelihood
9/26/2004 Stat 567: Unit 14 - Ramón León 42
Prediction of Future Events
DATA
22
9/26/2004 Stat 567: Unit 14 - Ramón León 43
Approximating Predictive Distributions
9/26/2004 Stat 567: Unit 14 - Ramón León 44
Location-Scale Based Prediction Problems
23
9/26/2004 Stat 567: Unit 14 - Ramón León 45
Predicting a New Observation
9/26/2004 Stat 567: Unit 14 - Ramón León 46
Predictive Density and Prediction Intervals for a Future Observation from the Bearing Cage Population
24
9/26/2004 Stat 567: Unit 14 - Ramón León 47
Caution on the Use of Prior Information• In many applications, engineers really have useful,
indisputable prior information. In such cases, the information should be integrated into the analysis
• We must beware of the use of wishful thinkingas prior information. The potential for generating serious misleading conclusions is high.
• As with other inferential methods, when using Bayesian methods, it is important to do sensitivity analysis with respect to uncertain inputs to ones model (including the inputted prior information
9/26/2004 Stat 567: Unit 14 - Ramón León 48
SPLIDA Bayes Analysis
25
9/26/2004 Stat 567: Unit 14 - Ramón León 49
SPLIDA Bayes Analysis:Bearing Cage Data
9/26/2004 Stat 567: Unit 14 - Ramón León 50
26
9/26/2004 Stat 567: Unit 14 - Ramón León 51
9/26/2004 Stat 567: Unit 14 - Ramón León 52
0.01 quantile 100 500 2000
123456
beta
5 6 7 8log(0.01 quantile)
2 3 4 5 6beta
Weibull Model Prior Distribution for BearingA data
log100 4.6log5000 8.51
==
27
9/26/2004 Stat 567: Unit 14 - Ramón León 53
9/26/2004 Stat 567: Unit 14 - Ramón León 54
0.01 quantile 100 200 500 1000 2000 5000
1.5
2.0
2.5
3.0
3.5
4.0
4.5
5.0
5.5
beta
Sun Sep 29 18:24:50 2002
0.010.1 0.5 0.9
Weibull Model Prior Distribution for Bearing Cage Data
28
9/26/2004 Stat 567: Unit 14 - Ramón León 55
9/26/2004 Stat 567: Unit 14 - Ramón León 56
0.01 quantile 500 1000 2000 4000
1.5
2.5
3.5
4.5
beta
6.5 7.0 7.5 8.0 8.5log(0.01 quantile)
1.5 2.5 3.5 4.5beta
Weibull Model Posterior Distribution for Bearing Cage Data
29
9/26/2004 Stat 567: Unit 14 - Ramón León 57
SPLIDA Posterior of B10
9/26/2004 Stat 567: Unit 14 - Ramón León 58
t_0.11500 2500 3500 4500 5500
f(t_0
.1 |
DA
TA)
Sun Sep 29 18:12:26 2002
Weibull Model Posterior Distribution for Bearing Cage Data
The mean of the posterior distribution of t_0.1 is: 2863The 95% Bayesian credibility interval is: [2011, 4361]
30
9/26/2004 Stat 567: Unit 14 - Ramón León 59
beta2.0 2.5 3.0 3.5 4.0
f(bet
a | D
ATA)
Sun Sep 29 18:12:26 2002
Weibull Model Posterior Distribution for Bearing Cage Data
The mean of the posterior distribution of beta is: 2.842 The 95% Bayesian credibility interval is: [2.184, 3.625]
9/26/2004 Stat 567: Unit 14 - Ramón León 60
31
9/26/2004 Stat 567: Unit 14 - Ramón León 61
0.1 quantile 100 500 2000 5000 20000
1.5
2.0
2.5
3.0
3.5
4.0
4.5
5.0
5.5
beta
Sun Sep 29 18:43:33 2002
0.010.10.50.9
Weibull Model Posterior Distribution for Bearing Cage Data
9/26/2004 Stat 567: Unit 14 - Ramón León 62
32
9/26/2004 Stat 567: Unit 14 - Ramón León 63
F(5000)0.0 0.2 0.4 0.6 0.8 1.0
f(F(5
000
| DA
TA)
Sun Sep 29 20:12:46 2002
Weibull Model Posterior Distribution for Bearing Cage Data
The mean of the posterior distribution of F(5000) is: 0.4589 The 95% Bayesian credibility interval is: [0.1355, 0.9071]