Thompson Lumber

1
Thompson Lumber
0.7
0.7
1
2
0.35
0.5
PosSurvey
PosSurvey
0.5
0.3
0.3
FM
FM
0.15
NegSurvey
NegSurvey
0.2
0.2
0.1
0.5
PosSurvey
PosSurvey
0.5
0.8
0.8
UM
UM
NegSurvey
0.4
NegSurvey
This tree came directly from the word problem on
page 126. For example, John guesses 50 chance
of a favorable market and experts say that with a
products with a favorable market were correctly
predicted 70 of the time. The rest is filled in
similarly.
Multiplying out the probability branches, you end
up with the total probability for a Positive
Survey and a Negative Survey equal to P(Positive
survey) 0.35 0.1 0 .45 P(Negative survey)
0.15 0.4 0.55
0.78
3
4
0.35
0.35
0.45
0.45
FM
FM
0.22
PosSurvey
PosSurvey
0.1
0.1
UM
UM
0.27
0.15
0.15
0.55
0.55
FM
FM
0.73
NegSurvey
NegSurvey
0.4
0.4
UM
UM
Redraw the tree using the new data (the survey
probabilities) and rearrange the final
probabilities to the appropriate tree branch
(e.g., the NegSurvey and FM still has the same
branch probability of 0.15.
At this point, you recalculate the FM and UM
probabilities by dividing each branch probability
by the probability of the survey branch. These
then are your baysian probabilities for a
Favorable Market vs. an Unfavorable Market based
on the survey results. For example P(FMPosSurvey
) 0.35/.045 0.78P(FM NegSurvey) 0.4/0.55
0.73
Help fromChris Mohr ltplease.zeei_at_verizon.netgt