Você está na página 1de 17

Birth months of a group of 94 students

Month

Jan

Feb

Mar

Apr

May

Jun

Jul

Aug

Sep

Oct

Nov

Dec

Number

10

10

11

10

15

Probability that birth month is April is 10/94 = 0.106

Birth months by gender


Month

Jan

Feb

Mar

Apr May Jun

Jul

Aug Sep

Oct

Nov Dec

Females

Males

Birth months- conditional probabilities

Month

Jan

Feb

Mar

Apr May Jun

Females

Males

Number of males = 54

Jul

Aug Sep

Oct

Nov Dec

Number of females = 40

April birth month: What if the person chosen is known to be male?

Probability is now 7/54 = 0.1296 = P(AprilMale)


Probability that birth month is April:
For a person: 0.106
For a male: 0.1296
For a female: 0.075

Restricting the sample space


S
C

If C is known to have occurred then how does P(A) change?


P(A) is the size of the red event relative to the entire sample space.
With the extra information that C has occurred, this changes to the
size of the common bit, relative to the size of C
for P(C)0

= P(A and C)
P(C)

Rules for tree diagrams


Probability entered on a branch is that of
reaching the end of that branch, conditional
on having reached the start

Probabilities emanating from any node must


add to 1 (i.e. all possibilities catered for)

Multiplication principle
Probability of traversing a path of branches
is the product of the probabilities on the
path.

Tossing three coins

Given that I throw at least one head, what


is the chance that I throw three?
Restricted sample space method

Head

Head

Head

We should ignore the bottom path,


leaving 7 equally likely outcomes.

Tail

Probability of getting three heads


is thus 1/7.

Head
Tail

Tail

Head

Head
Tail

Tail

Head
Tail

Tail

From definition of conditional probability


There are 8 equally likely outcomes

P(3 heads and at least 1 head)= P(3 heads)=1/8


P(at least one head) = 7/8
1/ 8 1
So P(3 H at least 1) =

7/8 7

Example 1
A survey of customers using credit cards in a home improvements store
showed that 60% were men (M).
Of these two thirds used a store card (C) of whom only one in ten had a
positive balance.
Of the women (W), three quarters had a store card and one third of these
had a positive balance.
Find the probability that a randomly chosen customer is
(i) a man with a credit card
(ii) a woman without a credit card
(iii) a woman with a positive balance
(iv) What proportion of customers who have a positive balance are
women?

Example 1
i) P(M and C) = 0.6 x 2/3
0.1
2/3

C
0.9

M
0.6

1/3

3/4

iv) P(W+) =P(W and +)

2/3

1/4

=0.4 x = 0.1
iii) P(W and +) = 0.4 x x 1/3 = 0.1

W
Not C

=0.4

Neg ii) P(W and no card)

Not C
1/3

0.4

Pos

Pos
Neg

P(+)

= 0.1/0.14 = 0.714
[ P(+) = 0.6 x 2/3 x 0.1 + 0.1 = 0.04 +
0.1 + 0.14 ]

Example 2
When recruiting for a management post, 30% of applicants are immediately rejected
on academic grounds; references are taken up on the remainder and only 5% of all
references are unsatisfactory. Those with satisfactory references are interviewed; 20%
of interviewees are rejected and the rest go on for final selection.
(i) What is the probability that a candidate goes onto final selection?
(ii) What proportion are rejected?
(i) P(Final selection) = 0.7 x 0.95 x 0.8 = 0.532
(ii) P(rejection) = 0.3 + 0.7 x 0.05 + 0.7 x 0.95 x 0.2 = 0.468 or 1 0.532

0.95
0.7

OK

Reject

0.8

Final
selection

0.2
0.05

0.3

Reference
OK

Reference
not OK

Rejected
at
interview

Medical diagnosis example


It is known that 10% of the population suffers from a
certain disease.
A diagnostic test for the disease gives a positive result
with probability 0.95 if the disease is present. If the
disease is not present, the probability of a positive
result is 0.02.
A test is given to a randomly selected person.
Find the probability of this test giving a positive result.
Given that the test gives a positive result, calculate
the probability that the person has the disease.

Medical diagnosis example


POSITIVE
DISEASED

NEGATIVE

POSITIVE
HEALTHY

NEGATIVE

Medical diagnosis example


POSITIVE
95%
DISEASED

5%

NEGATIVE
10%

POSITIVE
2%

90%
HEALTHY

98%

NEGATIVE
P( positive result) = 0.1 x 0.95 + 0.9 x 0.02 = 0.113

Medical diagnosis example


84% of those with positive results have the disease 16% of those with positive results
do NOT have the disease!

POSITIVE
95%
DISEASED

5%

NEGATIVE
10%

POSITIVE
2%

90%
HEALTHY

98%

NEGATIVE
P( has disease positive) =

0.1 x 0.95 = 0.84


0.113

Expected Value and Decision Trees

The Expected Value of a random quantity is


calculated by multiplying the outcome by its
probability.
Examples.
1. Value of sale = 150, Probability of sale = 2/3
Expected value =150 x 2/3 = 100
2. If there are two possibilities:
Profit = 200 with probability 0.6
Loss = -100 with probability 0.4
Expected profit = 200 x 0.6 -100 x 0.4 =120 - 40 = 80

3. Five possible outcomes:


Profit ()
-200 -100 0
100 200
Probability 0.1 0.2 0.25 0.35 0.1
Expected: -20 + - 20 + 0 + 35 + 20 = 15
So the expected profit is 15

Decision Trees.
Tree diagrams may also be used to show the
probability and expected value of an outcome under
different strategies.
Example
A company is considering three possible sales
strategies. If strategy A is a success the predicted
profit is 5 million as against an 8m loss if it is a
failure. The corresponding figures for strategy B
are: success 10m profit, failure 4m loss. For C
they are profit 5m, loss 5m. The probabilities of
success in each of A, B and C are respectively 0.8,
0.2 and 0.9.
Which strategy brings the greatest expected profit?

Solution
0.8

0.2
0.2

5 m

F
S

- 8 m
10 m

B
0.8
0.9

-4 m

5 m

-5 m

C
0.1

Strategy A
Expected profit = 0.8 x 5 0.2 x 8 = 2.4 million
Strategy B
Expected profit = 0.2 x 10 0.8 x 4 = -1.2 million
Strategy C
Expected profit = 0.9 x 5 0.1 x 5 = 4 million

Choose Strategy C

Você também pode gostar