1

Chapter 2 Basic Probability

10% defective

90% OK

Two possible outcomes:

Defective

Non-defective

For an item chosen at random:

P(defective) = 0.1

P(non-defective) = 0.9

i.e., P(defective) + P(non-defective) = 1

Basic Rule: Probabilities for all possible outcomes must sum to 1.

A more complex situation: two outcomes and two machines

Machine 1 Defect. Machine 2

0.1 0.5 0.6

0.1 0.3 0.4

0.2 0.8 1.0

Non-def.

Machine 1 produces 0.6 (60%) of the items Machine 2 produces 0.4 (40%) of the items 0.2 (20%) of the items are Defective 0.8 (80%) of the items are Non-defective

Total Probability: combine all possible events for a particular outcome Machine 1 Defect. Non-def. Machine 2

0.1

0.1

0.2

0.5
0.6

0.3
0.4

0.8
1.0

Law of total probability: P(defective) = P(defective & machine 1) + P(defective & machine 2)

Conditional Probability: for events that occur

when another event occurs

Machine 1 Defect. Non-def. Machine 2

0.1 0.5 0.6

0.1 0.3

0.2 0.8

0.4

1.0

Conditional probability: Given the item is from Machine 1 ... 1/6 defective

Conditional Probability: for events that occur

when another event occurs

Machine 1 Defect.
Non-def.

Machine 2

0.1 0.5 0.6

0.1 0.3
0.4

0.2 0.8
1.0

Conditional probability: notation P(defective | machine 1) = 1/6 given

Conditional Probability: for events that occur

when another event occurs

Machine 1 Defect.
Non-def.

Machine 2

0.1 0.5 0.6

0.1 0.3
0.4

0.2 0.8
1.0

P(machine 1 | defective) =

Conditional Probability: for events that occur

when another event occurs

Machine 1 Defect.
Non-def.

Machine 2

0.1 0.5

0.1 0.3 0.4

0.2 0.8
1.0

0.6
P(non-defective | machine 2) =

Compound Events: combination of two or more simple events Machine 1 Defect. Machine 2

0.1 0.5

0.1 0.3 0.4

0.2 0.8 1.0

Non-def.

0.6

Compound event: Defective & Machine 1

10

Compound Events: Probabilities

Machine 1 Defect.
Non-def.

Machine 2

0.1 0.5 0.6

0.1 0.3
0.4

0.2 0.8
1.0

Ingredients: Taking Machine first, P( Machine 1) = 0.6 Then, given Machine 1, P( Defective | machine 1)= 1/6

11

Compound Events: Probabilities P( Machine 1 & Defective) = P( Machine 1) P(Defective | Machine 1) = 0.6 1/6 = 0.1

Fundamental multiplication rule:

P(A & B) = P(A) P(B | A)

[Think about a sequence of events]

12

Example:

Birthday Problem (for 10 people)

P(at least two share a birthday)?

Too hard to do directly. Look at it another way

P(none share a birthday)?

13

Example:

Birthday Problem (for 10 people)

Look at it another way

P(none share a birthday)?

Any

364 365

363 365

362 365

356 365

14

Example:

Birthday Problem (for 10 people)

364 363 362 Any 1 365 365 365

356 365

= P(none share a birthday) = 0.883 1 0.883 = 0.117 P(at least 2 share a birthday) = ??

15

Example:

Birthday Problem (in general) P(at least 2 share a birthday)

No. of People

10
20 30 40 50 60

0.117
0.411 0.706 0.891 0.970 0.994

16

Example: Brand loyalty

1st car 60% 40%

2nd car Same brand Different

3rd car 90% 10% Same Different

P( three cars in a row of same brand) = ?? = P(2nd car same) P(3rd car same | 2nd car same) = 0.6 0.9 = 0.54

17

Example: Suppose 65% of people have a mobile phone, 25% have cable TV and 20% have both. What percentage have either? Phone Yes No Yes Cable No Total 0.65

Total

18

Example: Suppose 65% of people have a mobile phone, 25% have cable TV and 20% have both. What percentage have either? Phone Yes No Yes Cable No Total 0.65 Total 0.25

19

Example: Suppose 65% of people have a mobile phone, 25% have cable TV and 20% have both. What percentage have either? Phone Yes No 0.20 Total 0.25

Yes Cable No Total

0.65

20

Example: Suppose 65% of people have a mobile phone, 25% have cable TV and 20% have both. What percentage have either?

Yes 0.20 Cable No Total 0.45 0.65

Phone Yes No

0.05
0.30 0.35

Total 0.25 0.75 1.00

Answer = 0.20 + 0.05 + 0.45 = 0.70

21

Example: 75% of your customers are satisfied with the service and 60% are aged 50 years or more. One third of the customers who are satisfied are aged less than 50 years.

Satisfied Yes No
50 + Yes No Total 0.75

Total

22

Example: 75% of your customers are satisfied with the service and 60% are aged 50 years or more. One third of the customers who are satisfied are aged less than 50 years. Satisfied Yes No 50 + Yes No Total Total 0.60

0.75

23

Example: 75% of your customers are satisfied with the service and 60% are aged 50 years or more. One third of the customers who are satisfied are aged less than 50 years. Satisfied Yes No 50 + Yes No Total 0.25 Total 0.60

0.75

24

Example: 75% of your customers are satisfied with the service and 60% are aged 50 years or more. One third of the customers who are satisfied are aged less than 50 years. Satisfied Yes No 0.50 0.10 0.15 0.25 Total 0.60 0.40 1.00

50 +

Yes No Total 0.25

0.75

25

Example: A technical process has an 80% chance of succeeding if carried out by company A, but only a 60% chance if carried out by B. Company A is used 90% of the time. What is the probability overall that the process succeeds? Succeeds Yes No Total 0.9 0.1 1.0

A CompanyB Total

26

Example: A technical process has an 80% chance of succeeding if carried out by company A, but only a 60% chance if carried out by B. Company A is used 90% of the time. What is the probability overall that the process succeeds? Succeeds Yes No Total 0.9 0.1 1.0

A CompanyB Total

27

Example: A technical process has an 80% chance of succeeding if carried out by company A, but only a 60% chance if carried out by B. Company A is used 90% of the time. What is the probability overall that the process succeeds? Succeeds Yes No 0.72 0.18 0.04 Total 0.9 0.1 1.0

A CompanyB Total 0.06

28

Example: A technical process has an 80% chance of succeeding if carried out by company A, but only a 60% chance if carried out by B. Company A is used 90% of the time. What is the probability overall that the process succeeds? Succeeds Yes No 0.72 0.18 Total 0.9 0.1

A CompanyB Total 0.06

0.78

0.04
0.22

1.0

29

Mutually exclusive events - when 2 or more events cannot both (all) occur. Examples:

Gender: Female or male

Coin toss: head or tail Hair colour: red, black, brown, blonde, Satisfaction levels: poor, satisfactory, good Incomes: <$20,000, $20,000-$40,000, >$40,000

30

Mutually Exclusive probabilities may be added

<$20,000

$20,000-$40,000

>$40,000

Probs:

6/15

5/15

4/15

P(>$40,000 or <$20,000) = P(>$40,000) + P(<$20,000) = 4/15 + 6/15 = 10/15

31

(Statistical) Independence When occurrence of one outcome does NOT alter the chance of another.

Your Lotto ticket wins Your income exceeds $50,000

Independent P(Win | Income > $50,000) = P(Win)

32

(Statistical) Independence When occurrence of one outcome does NOT alter the chance of another. Toss a Head with a coin Last toss was a Tail

Independent P( Head | Tail ) = P(Head) = 1/2

33

(Statistical) Independence When occurrence of one outcome does NOT alter the chance of another. Person votes Yes in a referendum Persons sister votes Yes in the referendum

Independent? NO P( Yes | Sister votes Yes ) > P( Yes)?

34

Most data analysis requires that each observation in the data is INDEPENDENT of the others. Examples: Collect a sample of income data Survey questions at shopping centres Defective/non-defective items from a process Independent events: probability rules If A and B are independent events:

P(A and B) = P(A) P(B) P(A | B) = P(A) and P(B | A) = P(B)

35

Example

A process consists of 4 components which fail independently of each other.

A B C D

P(fail)

0.01

0.03

0.02

0.06

P(individual failure in any year) P(system failure in any year) = ??

36

Example
A

0.01

0.03

0.02

0.06

P(no failure) = P(A no fail) P(system failure in any year) ? no fail) P(C no fail) = P(B P(D no fail) = (1-0.01) (1-0.03) (1-0.02) (1-0.06) = 0.885

P(system failure) = 1 - 0.885 = 0.115