BCCS 2008/09: GM&CSS
description
Transcript of BCCS 2008/09: GM&CSS
![Page 1: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/1.jpg)
BCCS 2008/09: GM&CSS
Lecture 6:
Bayes(ian) Net(work)s and
Probabilistic Expert Systems
![Page 2: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/2.jpg)
A. Motivating examples
• Forensic genetics
• Expert systems in medical and engineering diagnosis
• Bayesian hierarchical models
• Simple applications of Bayes’ theorem
• Markov chains and random walks
![Page 3: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/3.jpg)
The ‘Asia’ (chest-clinic) example
The results of a single chest X-ray do not discriminate between lung cancer and tuberculosis, as neither does the presence or absence of dyspnoea.
A recent visit to Asia increases the risk of tuberculosis, while smoking is known to be a risk factor for both lung cancer and bronchitis.
Shortness-of-breath (dyspnoea) may be due to tuberculosis, lung cancer, bronchitis, more than one of these diseases or none of them.
+2
![Page 4: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/4.jpg)
Visual representation of the Asia example - a graphical model
![Page 5: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/5.jpg)
The ‘Asia’ (chest-clinic) example
Now … a patient presents with shortness-of-breath (dyspnoea) …. How can the physician use available tests (X-ray) and enquiries about the patient’s history (smoking, visits to Asia) to help to diagnose which, if any, of tuberculosis, lung cancer, or bronchitis is the patient probably suffering from?
![Page 6: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/6.jpg)
An example from forensic geneticsDNA profiling based on STR’s (single tandem
repeats) are finding many uses in forensics, for identifying suspects, deciding paternity, etc. Can we use Mendelian genetics and Bayes’ theorem to make probabilistic inference in such cases?
![Page 7: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/7.jpg)
Graphical model for a paternity enquiry - allowing mutation
Having observed the genotype of the child, mother and putative father, is the putative father the true father?
![Page 8: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/8.jpg)
Surgical rankings
• 12 hospitals carry out different numbers of a certain type of operation:
47, 148, 119, 810, 211, 196, 148, 215, 207, 97, 256, 360 respectively.
• They are differently successful, and there are:
0, 18, 8, 46, 8, 13, 9, 31, 14, 8, 29, 24 fatalities, respectively.
![Page 9: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/9.jpg)
Surgical rankings, continued
• What inference can we draw about the relative qualities of the hospitals based on these data?
• Does knowing the mortality at one hospital tell us anything at all about the other hospitals - that is, can we ‘pool’ information?
![Page 10: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/10.jpg)
B. Key ideas in exact probability calculation in complex systems• Graphical model (usually a directed acyclic
graph)
• Conditional independence graph
• Decomposability
• Probability propagation: ‘message-passing’
![Page 11: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/11.jpg)
A B C
A Conditional independence graph (CIG) has variables as nodes and (undirected) edges between pairs of nodes – absence of an edge between A and C means A C| (rest), e.g.
Conditional independence graphs
A B C
D E
![Page 12: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/12.jpg)
A B C
A B C
Directed acyclic graph (DAG)
… indicating that model is specified by p(C), p(B|C) and p(A|B): p(A,B,C) = p(A|B)p(B|C)p(C)
A corresponding Conditional independence graph (CIG) is
… encoding various conditional independence assumptions, e.g. p(A,C|B) = p(A|B)p(C|B)
![Page 13: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/13.jpg)
definition of p(C|B)true for any A, B, C
A B C
A B C
DAG
CIG
)|(),(),|(),(),,( BCpBApBACpBApCBAp
)(
),(),(
Bp
CBpBAp
since BAC |
+4
![Page 14: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/14.jpg)
A B CCIG
),,|(),|(),(),,,( CBADpBACpBApDCBAp
)()(
),(),(),(
BpBp
DBpCBpBAp
D
)|()|(),( BDpBCpBAp
+2
![Page 15: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/15.jpg)
A B CCIG
),,,|(),|,(),(),,,,( DCBAEpBADCpBApEDCBAp
),()(
),,(),,(),(
DCpBp
EDCpDCBpBAp
D
),|()|,(),( DCEpBDCpBAp
E
+2
![Page 16: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/16.jpg)
A B CCIG
),,,,( EDCBAp),()(
),,(),,(),(
DCpBp
EDCpDCBpBAp
D E
![Page 17: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/17.jpg)
A B CCIG
D E
AB BCDBJT CDECD
),,,,( EDCBAp),()(
),,(),,(),(
DCpBp
EDCpDCBpBAp
SseparatorsS
cliquesCC
Xp
Xp
)(
)(
+1
![Page 18: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/18.jpg)
An important concept in processing information through undirected graphs is decomposability
(= graph triangulated
= no chordless
-cycles)
Decomposability
7 6 5
2 3 41
4
![Page 19: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/19.jpg)
Cliques
A clique is a maximal complete subgraph: here the cliques are {1,2},{2,6,7}, {2,3,6}, and {3,4,5,6}
7 6 5
2 3 41
![Page 20: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/20.jpg)
7 6 5
2 3 41
12
267 236 345626 36
2
a separator
The running intersection property:For any 2 cliques C and D, CD is a subset of every node between them in the junction tree
A graph is decomposableif and only if it can be represented by ajunction tree (which isnot unique)
+4
another cliquea clique
![Page 21: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/21.jpg)
Non-uniquenessof junction tree
7 6 5
2 3 41
12
267 236 345626 36
2
![Page 22: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/22.jpg)
7 6 5
2 3 41
12
267 236 345626 36
2
12
2
Non-uniquenessof junction tree
![Page 23: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/23.jpg)
C. Exact probability calculation in complex systems
0. Start with a directed acyclic graph
1. Find corresponding Conditional Independence Graph
2. Ensure decomposability
3. Probability propagation: ‘message-passing’
![Page 24: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/24.jpg)
1. Finding an (undirected) conditional independence graph for a given DAG
• Step 1: moralise (parents must marry)
D E
CA B
F
D E
CA B
F
![Page 25: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/25.jpg)
1. Finding an (undirected) conditional independence graph for a given DAG
• Step 2: drop directions
D E
CA B
F
D E
CA B
F
D E
CA
B
F
![Page 26: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/26.jpg)
2. Ensuring decomposability
5 6 7
10 11
16
2
5 6 7
10 11
16
2
![Page 27: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/27.jpg)
2. Ensuring decomposability…. triangulate
5 6 7
10 11
16
2
5 6 7
10 11
16
2
5 6 7
10 11
16
2
![Page 28: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/28.jpg)
3. Probability propagation
5 6 7
10 11
16
2
2 5 6 7
5 6 7 11
5 6 10 11
10 11 16
5 6 7
5 6 11
10 11form junction tree
![Page 29: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/29.jpg)
SseparatorsS
cliquesCC
X
X
)(
)(
)(Xp
If the distribution p(X) has a decomposable CIG, then it can be written in the following potential representation form:
the individual terms are called potentials; the representation is not unique
![Page 30: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/30.jpg)
A B CDAG
2/31/3C=1
4/73/7C=0
B=1B=0
1/32/3B=1
1/43/4B=0
A=1A=0
.3C=1
.7C=0B|CA|B
p(A,B,C) = p(A|B)p(B|C)p(C)
Problem setup
Wish to find p(B|A=0) , p(C|A=0)
![Page 31: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/31.jpg)
AB BCB
A B C
A B C
DAG
CIG
JT
Transformation of graph
![Page 32: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/32.jpg)
AB BCB
A B C
2/31/3C=1
4/73/7C=0
B=1B=0
1/32/3B=1
1/43/4B=0
A=1A=0
.3C=1
.7C=0B|CA|B
1/32/3B=1
1/43/4B=0
A=1A=0
.2.4B=1
.1.3B=0
C=1C=0
Initialisation of potential representation
1B=1
1B=0
![Page 33: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/33.jpg)
SseparatorsS
cliquesCC
X
X
)(
)(
)(Xp
We now have a valid potential representation
)(
),(),(),,(
B
CBBACBAp
but individual potentials are not yet marginal distributions
![Page 34: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/34.jpg)
marginalise
.2.4B=1
.1.3B=0
C=1C=0
1/32/3B=1
1/43/4B=0
A=1A=0
AB BCB
A B C
Passing message from BC to AB (1)
1B=1
1B=0
.6B=1
.4B=0
1/3 .6/12/3 .6/1B=1
1/4 .4/13/4.4/1B=0
A=1A=0
multiply
![Page 35: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/35.jpg)
.2.4B=1
.1.3B=0
C=1C=0
.2.4B=1
.1.3B=0
A=1A=0
AB BCB
A B C
Passing message from BC to AB (2)
.6B=1
.4B=0
.6B=1
.4B=0
1/3 .6/12/3 .6/1B=1
1/4 .4/13/4.4/1B=0
A=1A=0
assign
![Page 36: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/36.jpg)
AB BCB
.2.4B=1
.1.3B=0
C=1C=0
.2.4B=1
.1.3B=0
A=1A=0
.6B=1
.4B=0
A B C
After equilibration - marginal tables
![Page 37: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/37.jpg)
SseparatorsS
cliquesCC
Xp
Xp
)(
)(
)(Xp
We now have a valid potential representation where individual potentials are marginals:
)(
),(),(),,(
Bp
CBpBApCBAp
![Page 38: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/38.jpg)
AB BCB
.2.4B=1
.1.3B=0
C=1C=0
0.4B=1
0.3B=0
A=1A=0
.6B=1
.4B=0
A B C
.4B=1
.3B=0
.2 .4/.6.4 .4/.6B=1
.1 .3/.4.3.3/.4B=0
C=1C=0
Propagating evidence (1)
![Page 39: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/39.jpg)
Propagating evidence (2)
AB BCB
.133.267B=1
.075.225B=0
C=1C=0
0.4B=1
0.3B=0
A=1A=0
.4B=1
.3B=0
A B C
.4B=1
.3B=0
.2 .4/.6.4 .4/.6B=1
.1 .3/.4.3.3/.4B=0
C=1C=0
![Page 40: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/40.jpg)
SseparatorsS
cliquesCC
X
X
)(
)(
)(Xp
We now have a valid potential representation
)(
),(),(),,(
B
CBBACBAp
})0{()( AXpX EEwhere
for any clique or separator E
![Page 41: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/41.jpg)
AB BCB
.133.267B=1
.075.225B=0
C=1C=0
0.4B=1
0.3B=0
A=1A=0
.4B=1
.3B=0
A B C
.298.702
.7.208.492
totalC=1C=0
.571.429
.7.4.3
totalB=1B=0
Propagating evidence (3)
![Page 42: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/42.jpg)
Scheduling messages
There are many valid schedules for passing messages, to ensure convergence to stability in a prescribed finite number of moves.
The easiest to describe uses an arbitrary root-clique, and first collects information from peripheral branches towards the root, and then distributes messages out again to the periphery
![Page 43: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/43.jpg)
Scheduling messages
rootroot
![Page 44: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/44.jpg)
Scheduling messages
rootroot
![Page 45: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/45.jpg)
Scheduling messages
When ‘evidence’ is introduced - the value set for a particular node, all that is needed to propagate this information through the graph is to pass messages out from that node.
![Page 46: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/46.jpg)
D. Applications
An example from forensic genetics
DNA profiling based on STR’s (single tandem repeats) are finding many uses in forensics, for identifying suspects, deciding paternity, etc. Can we use Mendelian genetics and Bayes’ theorem to make probabilistic inference in such cases?
![Page 47: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/47.jpg)
Graphical model for a paternity enquiry - neglecting mutation
Having observed the genotype of the child, mother and putative father, is the putative father the true father?
![Page 48: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/48.jpg)
Graphical model for a paternity enquiry - neglecting mutationHaving observed the genotype of the child, mother and putative father, is the putative father the true father?
Suppose we are looking at a gene with only 3 alleles - 10, 12 and ‘x’, with population frequencies 28.4%, 25.9%, 45.6% -the child is 10-12, the mother 10-10, the putative father 12-12
![Page 49: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/49.jpg)
Graphical model for a paternity enquiry - neglecting mutation
we’re 79.4% sure the putative father is the true father
![Page 50: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/50.jpg)
Surgical rankings
• 12 hospitals carry out different numbers of a certain type of operation:
47, 148, 119, 810, 211, 196, 148, 215, 207, 97, 256, 360 respectively.
• They are differently successful, and there are:
0, 18, 8, 46, 8, 13, 9, 31, 14, 8, 29, 24 fatalities, respectively.
![Page 51: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/51.jpg)
Surgical rankings, continued
• What inference can we draw about the relative qualities of the hospitals based on these data?
• A natural model is to say the number of deaths yi in hospital i has a Binomial distribution yi ~ Bin(ni,pi) where the ni are the numbers of operations, and it is the pi that we want to make inference about.
![Page 52: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/52.jpg)
Surgical rankings, continued
• How to model the pi?
• We do not want to assume they are all the same.
• But they are not necessarily `completely different'.
• In a Bayesian approach, we can say that the pi are random variables, drawn from a common distribution.
![Page 53: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/53.jpg)
Surgical rankings, continued
• Specifically, we could take
• If and are fixed numbers, then inference about pi only depends on yi (and ni, and ).
),(~1
log Betap
p
i
i
![Page 54: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/54.jpg)
Graph for surgical rankings
iy
in ip
![Page 55: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/55.jpg)
Surgical rankings, continued
• But don't you think that knowing that p1=0.08, say, would tell you something about p2?
• Putting prior distributions on and allows `borrowing strength' between data from different hospitals
![Page 56: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/56.jpg)
Surgical rankings - simplified
3 hospitals, p discrete, only one hyperparameter
![Page 57: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/57.jpg)
Surgical rankings - simplified
prior for prior for pi given
![Page 58: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/58.jpg)
Surgical rankings
![Page 59: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/59.jpg)
Surgical rankings
![Page 60: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/60.jpg)
Surgical rankings
![Page 61: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/61.jpg)
Surgical rankings
![Page 62: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/62.jpg)
The ‘Asia’ (chest-clinic) example
Shortness-of-breath (dyspnoea) may be due to tuberculosis, lung cancer, bronchitis, more than one of these diseases or none of them. A recent visit to Asia increases the risk of tuberculosis, while smoking is known to be a risk factor for both lung cancer and bronchitis. The results of a single chest X-ray do not discriminate between lung cancer and tuberculosis, as neither does the presence or absence of dyspnoea.
![Page 63: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/63.jpg)
Visual representation of the Asia example - a graphical model
![Page 64: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/64.jpg)
The ‘Asia’ (chest-clinic) example
Now … a patient presents with shortness-of-breath (dyspnoea) …. How can the physician use available tests (X-ray) and enquiries about the patient’s history (smoking, visits to Asia) to help to diagnose which, if any, of tuberculosis, lung cancer, or bronchitis is the patient probably suffering from?
![Page 65: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/65.jpg)
The ‘Asia’ (chest-clinic) example
![Page 66: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/66.jpg)
The ‘Asia’ (chest-clinic) examplequery('asia',c(0.01,0.99))query('smoke')tab(c('tb','asia'),,c(.05,.95,.01,.99),c('yes','no'))tab(c('cancer','smoke'),,c(.1,.9,.01,.99),c('yes','no'))tab(c('bronc','smoke'),,c(.6,.4,.3,.7),c('yes','no'))or('tbcanc','tb','cancer')tab(c('xray','tbcanc'),,c(.98,.02,.05,.95),c('yes','no'))tab(c('dysp','tbcanc','bronc'),,c(.9,.1,.8,.2,.7,.3,.1,.9),c('yes','no'))
prop.evid('asia','yes')prop.evid('dysp','yes')prop.evid('xray','no')pnmarg('cancer')
cancer=yes cancer=no 0.002550419 0.9974496
![Page 67: BCCS 2008/09: GM&CSS](https://reader036.fdocuments.in/reader036/viewer/2022062309/568152d9550346895dc0f4b9/html5/thumbnails/67.jpg)
Software
• The HUGIN system: freeware version
(Hugin Lite 5.7):
http://www.stats.bris.ac.uk/~peter/Hugin57.zip• Grappa (suite of R functions)
http://www.stats.bris.ac.uk/~peter/Grappa
7 6 5
2 3 41