Probability An Experiment is any action that leads to an uncertain outcome. A sample point is a simple outcome of an experiment. The sample space , denoted S is the set of all possible outcomes. Events are sets of outcomes or subsets of S and are denoted by capital letters such as A, B, … An event is simple if it consists of exactly 1 outcome. An event is compound if it consists of more than 1 outcome. Events = Sets Outcomes = Elements The order of elements in the set is not important. A = {1, 2, 3} = {3, 2, 1} {1, 3, 2} ,… Ex. Flip a fair coin twice. S = {HH, HT, TH, TT} Let B = event that exactly 1 H is flipped Let C = {at least 1 H is flipped} B = {HT, TH} = {TH, HT} C = {HH, HT, TH} Three basic set/event operations, Let E and F be two sets of S. 1. E compliment = E´ is the set of outcomes in S that are not in E. E´ is the event that E does not occur. 2. E intersection F = E ∩ F is the set of sample points that are in BOTH E and F or the set of sample points that are common to E and F. E ∩ F is the event that both event E and event F occur. 3. E union F = E U F is the set of sample points in E or in F (or in both). Any sample point that is in E is in E U F and any sample point that is in F is in E U F. the union is the set of outcomes that are in at least one of the events. E U F is the event that E occurs or F occurs. Note that E U F = F U E and E ∩ F = F ∩ E. and (E´)´= E The empty set or null set is the event that contains no outcomes. Denoted {} = Ø 1 When E ∩ F = {}, E and F are said to be disjoint or mutually exclusive. Example: Roll a fair die. So S = {1,2,3,4,5,6} Let A = {1, 3, 5} and B = {1, 2, 3, 4} Find A´, B´, A U B, A ∩ B. Draw Venn Diagram A´ ={2,4,6} A U B ={1,2,3,4,5} A´∩B ={2,4} (A ∩ B) ´ = {2,4,5,6} B´={5,6} A ∩ B ={1,3} A´UB ={1,2,3,4,6} (A U B)´= {6} Extension: let C = {2, 4} C´ = {1, 3, 5, 6} A ∩ C = {} A U C = {1, 2, 3, 4, 5} B ∩ C = {2, 4} B U C = {1, 2, 3, 4} B ∩ C´ = {1, 3} B U C´ = {1, 2, 3, 4, 5, 6} A ∩ C´ = {1, 3, 5} A U C´ = {1, 3, 5, 6} In general 2 sets (call them A and B) partition the sample space S into 4 regions. 2 I = A ∩B’ II = A ∩ B III = A’ ∩ B IV = A’ ∩ B’ Section 2.1 #2 3 vehicles exiting a freeway ramp each vehicle has 3 options, R, L, S Use T for sample space Before answering the questions, one should determine how big the sample space is and its elements (if feasible) Each car has 3 options and there cars, so the sample space will have 3 * 3 * 3 = 27 outcomes. 3 T={ RRR LLL SSS RRL LLR SSR RLR LRL SRS LRR RLL RSS RRS LLS SSL RSR LSL SLS SRR SLL LSS RLS LRS SRL RSL LSR SLR } a. A = {RRR, LLL, SSS} b. B = {RLS, LRS, SRL, RSL, LSR, SLR} c. C = {RRL, RLR, LRR, RRS, RSR, SRR} d. D = { RRL, RLR, LRR, RRS, RSR, SRR, LLR, LRL, RLL, LLS, LSL, SLL, SSR, SRS, RSS, SSL, SLS, LSS} e. D´ = {RRR, LLL, SSS, RLS, LRS, SRL, RSL, LSR, SLR} CUD=D C∩D=C De Morgan’s Laws. verify using Venn diagrams (A U B)´ = A´ ∩ B´ (A ∩ B)´ = A´ U B´ Section 2.2 Definitions of Probability. Simple: the probability of a sample point is the measure of the likelihood that outcome will occur when the experiment is run. The probability of an event, E, is sum of the probability of the sample contains contained in E. It is denoted P(E). You can think of P(E) in two ways: 1. If the experiment was run many many times, what fraction of the time would E occur. 2. If all the outcomes are equally likely to occur, then P(E) = number of ways that E occurs Total number outcomes of the experiment P(E) = number of outcomes in E number of outcomes in S 4 Axioms of Probability: 1. For all events E: 0 ≤ P(E) 2. P(S) = 1 3.a. If A1, A2, …, Ak are a finite set of Mutually exclusive events, then P(A1 U A2 U … Ak) = P(A1) + P(A2) + … + P(Ak) = ΣP(Ai) as i = 1, 2, … k 3.b. If A1, A2, … are an infinite set of Mutually exclusive events, then P(A1 U A2 U …) = P(A1) + P(A2) + … = ΣP(Ai) as i = 1, 2, … ∞ Note that Axioms 1 and 2 together mean the 0 ≤ P(E) ≤ 1 Back to the roll a fair die example: S = {1,2,3,4,5,6} A = {1, 3, 5} B = {1, 2, 3, 4} Find P(A), P(B), P(A U B), P(A ∩ B), P(A´), P(B´) A fair die means that the P(1) = P(2) =…=P(6) and since P(1) + P(2) + … + P(6) = 1, therefore if we let x = P(1), we have 6x = 1 so x = 1/6 Meaning that P(1) = 1/6, P(2) = 1/6, …, P(6) = 1/6 so since all the outcomes are mutually exclusive events: P(A) = P(1) + P(3) + P(5) = 3/6 = ½ = .5 P(B) = 4/6 = 2/3 = .667 P(A U B) = 5/6 = .833 P(A ∩ B) = 2/6 = 1/3 = .333 P(A´) = 3/6 = ½ = .5 P(B´) = 2/6 = 1/3 = .333 Looking at this example we can guess at 2 very important formulas for probability. 1. P(E´) = 1 – P(E) 2. P(E U F) = P(E) + P(F) – P(E ∩ F) (The additive rule of Probability) Note that in the above example P(A U B) = 5/6 = P(A) + P(B) – P(A ∩ B) = 3/6 + 4/6 – 2/6 P(B´) = 1/3 = 1 – 2/3 = 1 – P(B) 5 Proof that P(E´) = 1 – P(E) in general. E U E´ = S P(E U E´) = 1 P(E) + P(E´) – 0 = 1 P(E´) = 1 – P(E) In order to prove 2, note first that E U F = E U (F ∩ E´). Pf. (Show that E U F is a subset of E U (F ∩ E´) and then show E U (F ∩ E´) is a subset of E U F. a. Show that E U F is a subset of E U (F ∩ E´) Let x be an element of E U F, then x is in E or x is not in E. If x is in E then x is in E U (F ∩ E´). If x is not in E then x has to be in F, because x is in E U F, therefore x is in F ∩ E´, so x is in E U (F ∩ E´). Therefore E U F is a subset of E U (F ∩ E´). b. Show E U (F ∩ E´) is a subset of E U F. Let y be in E U (F ∩ E´), then y is in E or y is not in E. If y is in E, then y is in E U F. If y is not in E then y is in F ∩ E´, because y is in E U (F ∩ E´). Therefore y is in F so y is in E U F. Therefore E U (F ∩ E´) is a subset of E U F and E U (F ∩ E´) = E U F. Call this result * Also note that E and F ∩ E´ are mutually exclusive events. Finally note that for any 2 sets G and H, G = (G ∩ H) U (G ∩ H´) Proof: for you to do. Hint: draw a picture. Use argument similar to above. Call this result **. This means that P(G) = P(G∩H) + P(G ∩ H´) and P(G ∩ H´) = P(G) - P(G∩H). Call this result *** Prove P(E U F) = P(E) + P(F) – P(E ∩ F) Proof. P(E U F) = P(E) + P(F ∩ E´) by * and axiom 3. P(E U F) = P(E) + P(F) – P(F ∩ E) by ***. QED. An extension of this result to 3 sets is P(EUFUG) = P(E) + P(F) + P(G) – P(E∩F) – P(E∩G) – P(F∩G) +P(E∩F∩G). Draw Venn diagram 6 Labels: I = A ∩ B’ ∩ C’ III = A’ ∩ B ∩ C V=A∩B∩C VII = A’ ∩ B’∩ C II = A ∩ B ∩ C’ IV = A ∩ B’ ∩ C VI = A’ ∩ B ∩ C VIII = A’ ∩ B’∩ C’ Another way to calculate the P(E) for an event E: Let E = {a1, a2, …, ak} It may be easy to determine P(ai) for i = 1, 2, … , k P(E) = P(a1) + P(a2) + … + P(ak) Examples from Section 2.2: 12, 15, 18, 27 12. A = {student has a visa} B = {student has a master card} P(A) = .5, P(B) = .4 and P(A ∩ B) = .25 a. P(A U B) = .5 + .4 - .25 = .65 b. P(A´ ∩ B´) = P((A U B)´) = 1 - .65 = .35 c. P(A ∩ B´) = .5 - .25 = .25 15. Let A = {at most 1 purchase an electric dryer} This means 0 or 1 customers purchase electric. Since there are 5 people, the compliment of A is: A´ = { at least 2 people purchase an electric dryer} P(A´) = 1 - .428 = .572 b. P(all gas ) = .116 and P(all electric) = .005 P(at least one of each) = 1 - .116 - .005 = .879 18. 4 40W, 5 60W and 6 75W bulbs. P(at least 2 bulbs are selected to get one 75W) P(1st selected is not 75W) = 9/15 **** Added questions **** 7 1. Find the probability that the first 2 bulbs selected are 60W. 2. Find the probability that the first 2 bulbs selected are 60W and a 75W. Draw Tree diagram! 27. A = Anderson, B = Box, C = Cox, D = Cramer, F = Fisher Hint: A tree diagram may help determine S. Then list the outcomes, there are 10: 8 Section 2.3 Counting is essential to probability. Ex. How many ways can I order the letters A B C? ABC, ACB, BAC, BCA, CAB, CBA: which is 6. You can think of this as you have 3 choices for the first letter and 2 choices for the second letter and 1 choice for the 3 letter. 3*2*1 =6 when you say “and” you multiply. This is known as the multiplicative rule of counting. If the first part of an experiment can be done in n1 ways and the second part can be done in n2 ways, then the experiment can be done in n1 * n2 ways. This can be generalized to an experiment with any number of parts. If an experiment has k parts and the ith part can be done in ni ways (ni choices) as i = 1, 2, … , k then the entire experiment can be done in n1 * n2 * … nk ways. Menu Example: You own a restaurant and want to offer combination dinners. The patrons get a choice of 1 of 3 appetizers, 1 of 5 entrees and 1 of 2 desserts. How many combination meals are there? 3 * 5 * 2 = 30. Two Dice Example: 9 How many ways can I order the letters A B C D? 24 = 4 * 3 * 2 * 1 How many ways can I order n distinct letters (objects)? n! = n factorial. 0! = 1 n! = n * (n – 1)! for any positive integer n n! = n * (n-1) * (n-2) * …(2) * (1) 1! = 1 2! = 2 3! = 6 4! = 24 to find 5! On TI83/84, hit [5], [MATH], go over to PRB and the fourth option is 4:!(hit [4]). On screen you should see 5! Hit [ENTER] Answer is 120, which is 5 * 4! = 5 * 24 n! gets big really fast. n! = the number of ways to order n distinct items. Ex. How many ways can I select and order 2 letters from {A B C D E}? AB AC AD AE BC BD BE CD CE DE BA CA DA EA CB DB EB DC EC ED 20 ways = 5 * 4 This can also be looked at as a permutation. When you have n distinct objects and you select r of them and order is important, meaning AB ≠ BA, then you have n P r or P(r, n). n! n Pr (n r )! Notes: n will always ≥ r and n P r will reduce to n * (n – 1) * … * (n – r + 1) For the above example n = 5 and r = 2 so n P r = 5 P 2 = 5! / 3! = 120 / 6 = 20 Ex. How many ways can I select 2 letters from {A B C D E}? (here it assumed that order is not important) AB AC AD AE BC BD BE CD CE DE = 10 ways. 10 This can also be looked at as a combination. When you have n distinct objects and you select r of them and order is NOT important, meaning AB = BA, then you have n C r or C(r, n). Other Notation: nCr n! r!(n r )! In the above example: n = 5, r = 2 so n C r = 5 C 2 = 5! / (2! * 3!) = 120 / (2 * 6) = 10 nCn = 1 nC1 = n nC0 = 1 nC2 = n*(n-1)/2 Note: 6C0 = 1 6C1 = 6 6C2 = 15 6C3 = 20 6C4 =15 6C5 = 6 6C6 =1 Take the sum: 1+6+15+20+15+6+1 = 64 = 26. In Class exercises section 2.3 29, 32, 43, 44 43. 5 card poker. First find the ways to deal 5 cards: 52 C 5 = 2598960 How many types of straights are there? (A, 2, 3, 4, 5), (2, 3, 4, 5, 6), … (10, J, Q, K, A) So there are 10. How many ways can one get each of these? There are 4 of each type of card so 4 * 4 * 4 * 4 *4 = 45. Therefore P(straight) = 10 * 45 / 52 C 5 = 10240/2598960 = .004 Technically this is not 100% correct, why? A straight flush is better than a straight. How many way can one get a straight flush? 10 different types of straight flushes, but only 4 ways to get each. P(straight flush) = 40 / 52 C 5 = 40 / 2598960 = .00002 P(Straight) = 10200/2598960 = .004 11 Extra question: What is the probability of getting a flush? P(Flush) = 4 * 13 C 5 / 52 C 5 = 5148 / 2598960 = .002 Again you need to subtract P(Straight Flush) P(Flush) = 5108/2598960 = .002 Note that a flush beats a straight, because the probability of getting a flush is less than that of a straight. 12 The Partitions Rule: There are N distinct items. You wish to partition them into k sets. The ith set will contain ni items as i =1,2, …, k. For example the 1st set will contain n1 items. The number of distinct partitions is: N! n1!n2!...nk ! Note that n1 + n2 + … nk = N Ex. I have 15 students. How many ways can I break them up into 3 groups of 4, 5, and 6 students each? Unknown to me 2 of the students are sick, what is the probability that both sick students are in group of 6, assuming the students are randomly selected to the groups.. Ways: 15! /(4! 5! 6!) = 630630 P(Both in group of 6) = [13!/(4! 5! 4!)/630630] * 2!/2! = 90090/630630 = .143 Ex. How many ways can arrange the letters of the word LOOK? What makes this problem different is that the items are not distinct. Can you list them: However you can think of this as a partitions rule problem. You are partitioning the N (= 4) letters into L’s O’s and K’s. The answer is 4! /( 1! 2! 1!) = 12 Ex. How many ways can arrange the letters of the word MISSISSIPPI? 11!/(1! 4! 4! 2!) = 34650 Ex. My fan club has 10 members. 1. How many ways can I pick a President, Vice-President and Treasurer? 10 * 9 * 8 = 10 P 3 = 720 Order is important 2. How many ways can I pick 3 officers? 10 C 3 = 120 Order is not important 3. There are 6 males and 4 females in the club. If I pick 3 officers, what is the probability that they are all male? 13 6 C 3/ 10 C 3 = 20/120 4. For the same scenario as in 3, what is the probability that there is at least 1 female officer? 1 – 20/120 = 100/120 P(NONE) = 1 – P(AT LEAST 1) because {At least 1}’ = {None} 5. There are 6 males and 4 females in the club. If I pick 3 officers, what is the probability that exactly 2 are males and 1 is a female? (6 C 2) * (4 C 1) / (10 C 3) 44. Prove that n C k = n C (n – k) nCk n! (n k )!k! n! n! (n (n k ))!(n k )! k!(n k )! because n – (n – k) = k nC (n k ) 14 Section 2.4 Conditional Probability and Section 2.5 Independence Example One thousand people were asked to give their primary source of news. The results are given in the following table. A random subject is selected. Male Female Radio 275 200 TV 135 175 Paper 35 70 Other 60 50 Find the probability that a. the subject gets news from the Radio. P(Radio) = (275 + 200)/ 1000 = 475 / 1000 = .475 b. the subject gets news from Paper and is Female. P(Paper ∩ Female) = 70 / 1000 = .070 c. the subject gets news from TV or is Male P(TV U Male) = P(TV) + P(Male) – P(TV ∩ Male) = 310/1000 + 505/1000 – 135/1000 = 680/1000 = 0.680 d. the subject gets news from Paper GIVEN she is Female. This is a new type of probability called Conditional Probability. We are given additional information here, that the subject is Female. So we are limiting our Sample Space. P(Paper | Female) = 70 / 495 = .141 In general P(A | B) = P(A ∩ B) P(B) Provided that the P(B) ≠0 for the last example P(Paper ∩ Female) = 70/1000 = .07 and the P(Female) = 495/1000 = .495 so the P(Paper | Female) = .07 / .495 = .141 e. Find the probability that the subject is Female given she gets news from Paper. P(Female | Paper) = .07 / .105 = .667 = 70 / 105 Notes In general P(A | B ) ≠ P(B | A) whereas P(A ∩ B) = P(B ∩ A) and P( A U B ) = P( B U A) 15 P(A | B) is a probability so it is between 0 and 1, if you get a number bigger than 1 it means you made a mistake! Fix it! You may give answers as fractions or decimals but not both. Ex. Let P(A) = .32 and P(B) = .38 and P(A U B) = .55. Find the P(A | B) and P(B | A). First you need to find P(A ∩ B). P(A ∩ B) = .15 Why? P(A | B) = .15 / .38 = .395 P(B | A) = .15 / .32 = .469 Multiplicative Rule of Probability P(A ∩ B) = P(A | B) P(B) If you knew P(A | B) = .42 and P(B) = .27 then P(A ∩ B) = .42 * .27 = .1134 Two events (A and B) are independent when the fact that A occurs does not affect the probability that B will occur and vice versa. Equivalent definitions of Independent events A and B: P( A ∩ B ) = P(A) P(B) P(A | B) = P(A) P(B | A ) = P(B) So P(A ∩ B) = P(A) P(B) only if A and B are independent. Ex. the experiment is to draw a single card from a deck of 52. Let A be the event that the card is a Jack and let B be the event that the card is Black. Find the probability of A, B and A ∩ B, A|B, B|A ? Are A and B independent? P(A) = 4/52 = 1/13 P(B) = 26/52 = ½ P(A ∩ B) = 2/52 = 1/26 P(A | B) = 2/26 = 1/13 P(B | A) = 2/4 = ½ A and B are independent because: P(A ∩ B) = 1/26 = 1/13 * ½ = P(A) P(B) P(A | B) = 1/13 = P(A) P(B |A) = ½ = P(B) 16 Ex. Roll a fair die. Let A = {1, 3, 5} C = {1, 2, 3} Are A and C independent? P(A) = ½ P(C) = ½ P(A ∩ C) = 2/6 ≠ ½ * ½ = ¼ so A and C are not independent. P(A | C ) = 2/3 ≠ P(A) P(C | A) = 2/3 ≠ P(C) Ex. 2 fair dice are rolled. Let A = {Sum is odd} B = {sum > 9} C = {at least one die shows a 4} Recall that: P(A) = 18/36 = ½ P(B) = 6/36 = 1/6 P(C) = 11/36 P(A ∩ B) = 2/36 = 1/18 P(A ∩ C) = 6/36 P(B∩C) = 2/36 = 1/18 Now find P(A | B), P(B | A), P(A | C), P(C | A), P(B | C), P(C | B), P(A|B) = (2/36) / (6/36) = 2/6 = 1/3 P(B|A) = (2/36) / (18/36) = 2/18 = 1/9 P(A|C) = (6/36) / (11/36) = 6/11 P(C|A) = (6/36) / (18/36) = 6/18 = 1/3 P(C|B) = (2/36) / (6/36) = 2/6 = 1/3 P(B|C) = (2/36) / (11/36) = 2/11 ** DO not confuse Independent with Mutually Exclusive Events! if A and B are independent then P(A ∩ B) = P(A) P(B) if A and B are mutually exclusive then P(A ∩ B) = 0 Diagnostic testing. Let S = {condition or disease is present} Then S´ = {condition or disease is NOT present} Let T = { test for S is positive} Then T´ = { test for S is negative} Sensitivity = P(T | S ) Sensitivity = probability that the test is + given we know the subject has the condition Specificity = P(T´ | S´) Specificity = probability that the test is - given we know the subject does not have the condition P(False Positive) = P(T | S´) P(False Negative) = P(T´ | S) 17 Note that P(False Positive) = P(T | S´) = 1 – P(T´ | S´) and P(False Negative) = P(T´ | S) = 1 – P(T | S) Prevalence of a condition or disease is the percent or probability of the population that has it. Prevalence = P(S) What is usually of interest is the probability the subject actually has the condition given the test is + or the probability the subject actually does not have the condition given the test -. Tree Diagram ---P(T |S) ---------P(S)---| | ---P(T´|S) | | | ---P(T| S´) ---------P(S´)---| ---P(T´| S´) From the tree diagram we can calculate the intersection probabilities. 18 ---P(T | S) ---------P(S)---| | ---P(T´|S) | | | ---P(T|S´) ---------P(S´)---| ---P(T´|S´) P(T ∩ S ) = P(T|S) * P(S) P(T´ ∩ S ) = P(T´|S) * P(S) P(T ∩ S´ ) = P(T|S´) * P(S´) P(T´ ∩ S´) = P(T´|S´) * P(S´) Note that P(S) = P(S ∩ T) + P(S ∩ T´) and P(T) = P(T ∩ S) + P(T ∩ S´) Another methodology is to use Bayes’s rule Let A be any set. Then A = (A ∩ B) U (A ∩ B´). P(A) = P[(A ∩ B) U (A ∩ B´)] = P(A ∩ B) + P (A ∩ B´) – 0 P(A) = P(A | B) P(B) + P(A | B´) P(B´) (from Multiplicative Rule) P(B | A) = P(A ∩ B)/ P(A) and P(A ∩ B) = P(A |B) P(B) so P( A | B) P( B) P( B | A) P( A | B) P( B) P( A | B' ) P( B' ) It gives you a way to reverse the order of the conditional. Ex. It is known that about 1% of women who get a mammogram have breast cancer. The sensitivity of a mammogram is 0.86 and the specificity is 0.88. Real Questions: Given that the mammogram is +, what is the probability the woman has breast cancer? Given that the mammogram is -, what is the probability the woman does not have breast cancer? To get these answers we will make a tree diagram. We will then find P(T), P(T´), P(S ∩ T) and P(S´ ∩ T´) These will then be used to find P(S | T) and P(S´ | T´). Note what we are given in the problem: P(S) = .01, P(T|S) = .86 and P(T´|S´) = .88 which means: P(S´) = .99, P(T´|S) = .14 and P(T| S´) = .12 19 P(T|S)=.86 P(T ∩ S ) = .01 * .86 = .0086 ---P(S) = .01 | | P(T´|S)=.14 P(T´ ∩ S ) = .01 * .14 = .0014 | | | P(T|S´)=.12 P(T ∩ S´ ) = .99 * .12 = .1188 --P(S´) = .99 | P(T´|S´)=.88 P(T´ ∩ S´) = .99 * .88 = .8712 P(T) = P(T ∩ S) + P(T ∩ S´) P(T) = .0086 + .1188 = .1274 P(T´) = P(T´ ∩ S) + P(T´ ∩ S´) P(T´) = .0014 + .8712 = .8726 = 1 – P(T) To answer the questions: P(S | T) = P(T ∩ S) / P(T) = .0086 / .1274 = .0675 = 6.75% P(S´ | T´) = P(T´ ∩ S´) / P(T´) = .8712 / .8726 = .9984 = 99.84% Using Bayes’s Rule: P(S) = .01 P(T | S) = .86 P(T´ | S´) = .88 we want P(S | T) and P(S´ |T´) P( S | T ) P(T | S ) P( S ) .86 * .01 .0086 P(T | S ) P( S ) P(T | S ' ) P( S ' ) .86 * .01 .12 * .99 .1274 P(S | T) = .0675 by the same reasoning P(S´|T´) = .9984 The generalized version for Bayes’s Rule is given in your book: Given k mutually exclusive and exhaustive events: A1, A2, …, Ak such that P(A1) + P(A2) + … +P(Ak) = 1 (which means that A1, …, Ak form a partition of the sample space) and an observed event B, then P( Ai B) P( Ai) P( B | Ai) P( Ai | B) P( B) P( A1) P( B | A1) ... P( Ak ) P( B | Ak ) 20 Ex. A construction company employs 3 engineers, E1, E2 and E3. They use E1 for 30% of their jobs, E2 for 20% and E3 for 50%. The following probabilities are given: P(Error | E1) = .01, P(Error | E2) = .03, P(Error | E3) = .02,. If an error was made find the probability that it was made by E1. If an error was made find the probability that it was made by E2. If an error was made find the probability that it was made by E3. If an error was made who was most likely to make it? The denominator will be the same for all 3 probabilities. P(E|E1)P(E1) + P(E|E2)P(E2) + P(E|E3)P(E3) .01*.3 + .03*.2 + .02*.5 = .019 P(E1 | E) = P(E|E1)P(E1) / .019 = .003/.019 = .158 P(E2 | E) = P(E|E2)P(E2) / .019 = .006/.019 = .316 P(E3 | E) = P(E|E3)P(E3) / .019 = .010/.019 = .526 Events A1, …, An are mutually independent if for every k (k = 2, 3, …, n) and every subset i1, i2, …, ik, P(Ai1 ∩ Ai2 ∩ … ∩ Aik) = P(Ai1) * P(Ai2) * …* P(Aik) Note that this is a much stronger condition than pairwise independent. Ex. Roll 2 fair dice. one is red and one is green. A={red = 3} B={green = 4} C = {sum = 7} Are these events pairwise independent? Are these events mutually independent? P(A) = 1/6 P(B) = 1/6 P(C) = 1/6 Note 1/6 * 1/6 = 1/36 P(A ∩B) = 1/36 P(A ∩C) = 1/36 P(B∩C) = 1/36 Therefore the 3 events are pairwise independent But, P(A∩B∩ C) = 1/36 ≠ 1/6* 1/6 * 1/6 = 1/216 so they are not mutually independent. 21