Additive Rule/ Contingency Table
-
Upload
dieter-santana -
Category
Documents
-
view
14 -
download
0
description
Transcript of Additive Rule/ Contingency Table
JMB Ch2 Lecture 2 vSp2012 EGR 252.001 2012 Slide 1
Additive Rule/ Contingency TableExperiment: Draw 1 card from a standard 52 card deck. Record Value (A-K), Color & Suit. The probabilities associated with drawing an ace and with
drawing a black card are shown in the following contingency table:
Event A = ace Event B = black card Therefore the probability of drawing an ace or a black card is:
Type
Color
TotalRed Black
Ace 2 2 4
Non-Ace 24 24 48
Total 26 26 52
52
28
52
2
52
26
52
4)()()()( BAPBPAPBAP
JMB Ch2 Lecture 2 vSp2012 EGR 252.001 2012 Slide 2
Short Circuit Example - Data An appliance manufacturer has learned of an increased
incidence of short circuits and fires in a line of ranges sold over a 5 month period. A review of the defect data indicates the probabilities that if a short circuit occurs, it will be at any one of several locations is as follows:
The sum of the probabilities equals _____
Location P
House Junction (HJ) 0.46
Oven/MW junction (OM) 0.14
Thermostat (T) 0.09
Oven coil (OC) 0.24
Electronic controls (EC) 0.07
JMB Ch2 Lecture 2 vSp2012 EGR 252.001 2012 Slide 3
Short Circuit Example - Probabilities
If we are told that the probabilities represent mutually exclusive events, we can calculate the following:
The probability that the short circuit does not occur at the house junction is
P(HJ’) = 1 - P(HJ) = 1 – 0.46 = 0.54
The probability that the short circuit occurs at either the Oven/MW junction or the oven coil is
P(OM U OC) = P(OM)+P(OC) = 0.14 + 0.24 = 0.38
JMB Ch2 Lecture 2 vSp2012 EGR 252.001 2012 Slide 4
Conditional ProbabilityThe conditional probability of B given A is
denoted by P(B|A) and is calculated by
P(B|A) = P(B ∩ A) / P(A)Example:
S = {1,2,3,4,5,6,7,8,9,11} Event A = number greater than 6 P(A) = 4/10Event B = odd number P(B) = 6/10 (B∩A) = {7, 9, 11} P (B∩A) = 3/10
P(B|A) = P(B ∩ A) / P(A) = (3/10) / (4/10) = 3/4
JMB Ch2 Lecture 2 vSp2012 EGR 252.001 2012 Slide 5
Multiplicative RuleIf in an experiment the events A and B can both
occur, then
P(B ∩ A) = P(A) * P(B|A)Previous Example:
S = {1,2,3,4,5,6,7,8,9,11} Event A = number greater than 6 P(A) = 4/10Event B = odd number P(B) = 6/10P(B|A) = 3/4 (calculated in previous slide)
P(B∩A) = P(A)*P(B|A) = (4/10)*(3/4) = 3/10
JMB Ch2 Lecture 2 vSp2012 EGR 252.001 2012 Slide 6
Independence Definitions
If in an experiment the conditional probabilities P(A|B) and P(B|A) exist, the events A and B are independent if and only if
P(A|B) = P(A) or P(B|A) = P(B)
Two events A and B are independent
if and only if P (A ∩ B) = P(A) P(B)
JMB Ch2 Lecture 2 vSp2012 EGR 252.001 2012 Slide 7
Independence Example A quality engineer collected the following data on 100 defective
items produced by a manufacturer in the southeast:
What is the probability that the defective items were associated with the day shift? P(Day) = (20+15+25) / 100 = .60 or 60%
What was the relative frequency of defectives categorized as electrical? (20 + 10) / 100 P(Electrical) = .30
Are Electrical and Day independent? P(E ∩ D) = 20 / 100 = .20 P(D) P(E) = (.60) (.30) = .18 Since .20 ≠.18, Day and Electrical are not independent.
Problem/Shift Electrical Mechanical Other
Day 20 15 25
Night 10 20 10
JMB Ch2 Lecture 2 vSp2012 EGR 252.001 2012 Slide 8
Serial and Parallel Systems For increased safety and reliability, systems are often designed with
redundancies. A typical system might look like the following:
Principles:
0.95 0.9
0.88
0.85
0.97
A B
C
D
E
If components are in serial (e.g., A & B), all must work in order for the system to work.
If components are in parallel, the system works if any of the components work.
JMB Ch2 Lecture 2 vSp2012 EGR 252.001 2012 Slide 9
Serial and Parallel Systems
What is the probability that: Segment 1 works? A and B in series
P(A∩B) = P(A) * P(B) = (0.95)(0.9) = 0.855
Segment 2 works? C and D in parallel will work unless both C and D do not function
1 – P(C’) * P(D’) = 1 – (0.12) * (0.15) = 1-0.018 = 0.982
The entire system works? Segment 1, Segment 2 and E in series
P(Segment1) * P(Segment2) * P(E) = 0.855*0.982*0.97 =0.814
0.95 0.9
0.88
0.85
0.97
A B
C
D
E
12