Scribd Logo
Upload

Welcome to Scribd!

  • Chapter 7 - Rothenberg

    Uploaded by

    Kishawn Smith
    7 views25 pages

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    Report this Document

    Copyright:

    © All Rights Reserved
    Download as PDF or read online from Scribd
    Flag for inappropriate content
    7 views25 pages

    Chapter 7 - Rothenberg

    Uploaded by

    Kishawn Smith

    Copyright:

    © All Rights Reserved
    Download as PDF or read online from Scribd
    Flag for inappropriate content
    of 25
    DISCRETE PROBABILITY PROBLEMS AND BAYES’ FORMULA J) Elementary Probability Applications @ Total Probability Theorem and Bayes’ Formula 7-1. Elementary Probability Applications A, The two-stage approach We shall often make use of a two-stage approach for determining the prob- ability of an event, E. ‘Stage One [Or/Addition] First, we express the event £ as an “or” statement [equivalently, a “union” statement] which is in terms of mutually exclusive events Ay, Azy+o.s Axe That is, E= Ay or Azor...0r Ay (ay ‘Then, we take the probability of both sides of Eq. (1a), where on the right side we apply the addition rule. [See Axiom 1.3] Thus, we obtain Eq, (1b). P(E) = P(A\) + P(A) + + PUA) (aby ‘Stage Two [And/Multiplication] Next, we express each 4, as an “and” statement [equivalently, an “interseotion” statement] in terms of sequentially ‘occurring events such as By, Bs,..., By. That is, Ay = B, and B, and B, and... and B, (2a) ‘Then, we take the probability of both sides of Eq. (2a), where on the right side we apply the general multiplication rule [see Theorem 3.3]. Thus, we obtain Eg. (2b) P(A) = P(B,) P(Bs | By): PCBs | By and B,) 20) Now, Eq. (2b) reduces to the following equation if the events B,, B, Bs, Bare independent. (See Definition 3.3} P(A) = P(B,)- PUB): PCBs) PCB) (Qe) (This will be the case, for example, if By, By, By,-... B, are events which are respectively determined by r independent trials 1,2, ...,. Refer to Definitions 33 and 5.4 and Theorems 5.2 and 53 for a review of the concepts of independent trials and of events determined by independent trials. A key point is that such events are then independent events, in accordance with Definitions 3.2, 3.3, and 3.4] ‘The second stage thus consists of expressing each P(4,) probability by ‘an equation of the type (2b) or (20). Finally, we solve for P(E) by substituting the expressions thus obtained for P(A} P(A2).--. P(d,) back into Eq.(1b). Iss 156 Probability and Statitis ee SFSsSseseseF EXAMPLE 7-1: Tom wants to go on a date Saturday with either Alice or Barbara, He estimates (subjectively] that if he asks for a date, the probability that Alice will say yes is, while the probability Barbara will say yes is 54. Tom 23], the probability that at least two people will have the same birthday is larger than 5. For 60 ot more people in a group [r = 60] itis virtually certain that two or more people will have the same birthday. C, The general addition rule ‘The addition rules pertaining to P(A UB) and P(A W Bu C) were given in Theorems 2.1 and 2.2. The generalization for P(A, UAzU AsU...U Ay) where n is any positive integer, is stated as follows: “Theorem 7.1 (General Addition Rule): PUA, UA, U Ay UU) = (Ay) + PCAa) + PCA) [P(A Ay) + PA, Aa) 162 Probability and Statistics tet PALO As) + P(ALO As) et PAG OAD + [PLANO Ag) + PALO AOA) + $ (DPA, Ag a. AG) TThe subscript pairs in the P(4, 9 A,) terms constitute all possible con binations ofm things taken 2at a time [also, note that i /]. Thus, there are (() se ers The aubcrn ios in the PUA, 4) ems constitute {ips combinations of things akon 3 sng. (Ao, not that. and arealldiferent numbers Ths thereare (suche andson EXAMPLE 7-7: Write out Theorem 7.1 for the case where Solution PUA, UA As AQ) P(A,) + P(A) + P(As) + PlAg) = [P(A, A) + (AG Ag) + PAL AQ) + P(A, Ag) + P(A Aa) + PLAS Aad] $+ [P(A, ALO Ay) + PALA Og) + P(A, Ag Aa) + P(42 Ag Aa) = P(A AQ ASQ) 3 of the P(A, 4,0 Ay) type. The subscript pairs in the former terms, namely 12, 13, 14, 23, 24, and 34, constitute the six combinations of the four integers 1, 2, 3, and 4 taken two ata time, Likewise, the subscript trios in the latter terms, namely 123, 124, 134, and 234, constitute the four combinations of the four integers 1, 2, 3,and 4 taken three at a time, : A clear discussion which leads o a result similar to Theorem 7.1 appears in Chapter Five of Niven (1965). EXAMPLE 7-8: Given four chips numbered 1, 2, 3, and 4, which are in a container, and four envelopes labeled, respectively, with the numbers 1,2,3, and 4. The chips are drawn out one at a time from the container and placed in the envelopes, such thatthe first chip drawn out goes to envelope 1, the second chip t sim P(A Ay)=1 3 co) A permutation ofthe integers 1,2,3,...,7in which none ofthe n integers is in its natural place is called a derangement of the integers. Thus, for ‘example, the permutation 3412 is a derangement of the integers 1, 23,4. A question dealing with derangements occurs in Problem 7-3. The topics of desangements and the inclusion-exclusion principle [which is closely related to Theorem 7.1, the general addition rule] are covered in Niven (1965). 7-2. Total Probability Theorem and Bayes’ Formula ‘Suppose we havea sample space S, which is partitioned into a collection of events. [The latter events are, of course, mutually exclusive.] Our goal will be to develop ‘an equation for P(E) for some event F of § by making use of information dealing with the partition, A. Theorem of total probability Suppose that sample space $ is partitioned into k events By, Bry Bay «+s By. [That is, $= B, VB, UB, v...UB, and BoB, = @ fori}, Fora review of the partition concept, see Definition 1.8.) We also suppose that the probabilities P(B,), P(B,), P(B)--.. P(B,) are given or can easily be determined. For example, we demonstrate the key ideas for the case where we have 4 partition of $ into three events; that is, k = 3. Refer to Figure 7-3, where the three mutually exclusive events are B,, Ba, and By Thus, in Figure 7-3a, we have S=BLUB,UBy where BB, = B,OBy=BOB,=B (l) Now, suppose that E is any event [subset] of sample space S, as shown in Figure 7-3, Thus, E can be expressed as a union of three mutually exclusive events as follows: E By E)U(B, 9 E(B) OB) @ rote: In specific cases, one or mote of the (B, VE) terms may be the empty set. Now if say, (By \E) = 5, that means that E doesn’t intersect the subset By From the addition rule for mutually exclusive events [see, for example, ‘Axiom 1.37], we have P(E) = P(B, 9 E) + PUB, 0B) + PROB) eo Now, if we know [or can easily determine] the conditional probabilities P(E | B) fori = 1,2,and 3, then we can use the multiplication rule [Theorem 3.1] to write PUB E) = P\B)P{ESB) for i= 1,2, and 3 co} Finally, if we substitute from Eq.(4)foré = 1,2,and 3 into Eq, (3), weobtain P(E) = P(B,)PCE | By) + PUB)PUE | B,) + P(Bs)P(E| Bs) (8) ‘The generalization of Eq, (5) is known as the Theorem [or Law] of Total Probability. ‘Theorem 7.2 (Theorem of Total Probability: Ifthe sample space 5 is parti- tioned into k events By, Br, 5 By [hat is, S= By BU... By where 3.0 B, = @ for i # jJ, and.& is an event of the sample space S, then the probability of the event can be expressed as follows: 4 % 5 @ NS ane | one | one ) Figure 7-3. Diagrams for the Total Probability Theorem {for k= 3] 166 Probability and Statistics P(E) = P(B,)PCE | By) + P(B3)P(E | Ba) +... + P(B)PCE | By) Examples 7-10 and 7-11 illustrate Theorem 7.2 for cases in which the sample space is partitioned into two mutually exclusive events [i.c,, k = 2}. note: Actually, we have already applied the total probability theorem on several occasions in this book, without referring to it by name. For example, consider Examples 7-1 and 7-3 EXAMPLE 7-10: Suppose we choose two cards without replacement from a standard deck of 52 cards, Find the probability that the second card isa spade; denote the probability ofthis event by P(S2). (Note: We solved a similar problem in part (c) of Example 3-17.) Solution: Let's try to use common sense here, instead of just plugging numbers into an equation. Spade on card 2 can arise ifcard 1 is a spade [S,] and card 2 isa spade or ifcard | is a nonspade [S; ] and card 2is a spade. Thus, in symbols, ‘where S, indicates spade on card 2, we have $8, = (S, and 8;) oF (Sj and $,) = (8, 9) US, 9S;) wo ‘The two events on the right of Ea. (1) which are enclosed in parentheses are mutually exclusive, and so P(S,) = PIS, 9S;) + PS, OS;) @ ‘Then, applying the multiplication rule to the two terms on the tight of Eq. (2), wwe obtain P(S,) = PUS,)P(Sa | Sy) + POS)IP(S2 1 Si) @ [Note that Eq, (3) is a special case of Theorem 7.2 for k = 2. Here, Eis S,, B, is Sy, and By is S.J Now, the probability of spade on card | (ic, of P(S)] is £5. and the prbhitiyfnonpaleon cr feo 8) Teter 5). (3~1)_ 12 py cect dase gare GS ar Simlatly PU | Si) = ayy = spr sine® the cond tion here is for a nonspade on card 1. (For an iniportant relevant refer to Example 5-24,} Substituting these probability values into the of Eq. (3) eads to POS.) = GD) + GDGED = DED + GN = Ae = 2 o ‘Thus, we have obtained the surprising result that the unconditional prob- ability of spade on card 2 is 4, even though we have been drawing without replacement. In fact, it can be shown that i three, four, or more cards were drawn ‘without replacement, then the probability of spade on the last card would stil bbe sit is just as though only one card were drawn, is equal to EXAMPLE 7-11: Suppose a box contains three coins, of which two are fair coins and one is a coin with two heads. A coin is selected at random from the box and tossed once. What is the probability that the result of that tossisa head? ‘Solution: Again, et’s use a commonsense approach in which we express the event of interest [namely, head] in terms of two or more mutually exclusive events, Head [#1] can arise if the coin drawn is fair and it yields a head upon tossing or if'the coin drawn is two-headed and it yields a head upon tossing. Thus, denoting. the two-headed “bad” coin by B and a fair coin by F, we have Discrete Probability Problems and Bayes’ Formula 1H = (P and Hor (Band H)= (Fa H)U(BOH) aw ‘The two events on the right separated by the U are mutually exclusive. Thus, by the addition rule, we get P(H) = P(E H) + PB) @ Now, using the multiplication rule forthe two probabilities on the right, we get P(H) = POR)P(H | F) + P(B)P(H | BY @ Now, since there are two fair coins and one bad coin in the box, we know that POF) + j and P(B) = }. We also know that P(H | F) = § and PUT| B) = 1 ‘Thus, substituting these values into Eq. (3), we get PH) = A+ DO = 3 o Example 7-12 can be worked ovt by using a commonsense approach o by making use of Theorem 7.2 for k = 3 EXAMPLE 7-12: The Great Idea [GI] Company manufactures a certain type of light bulb at factories 1, I, and III, GI has in stock a large batch of bulbs, 25% of which were made ‘in factory 1, 35% in factory II, and 40% in factory THI. Also, the GI track record shows that 3% of all bulbs produced by factory I are detective, 5% of all bulbs produced by factory Hare defective, and 4% of all bulbs produced by factory III are defective. If single light bulb is picked at random from the batch, what is the probability that it is defective? Solution: Let's solve the problem by using the equation of Theorem 72for k = 3 Let E denote the event that “the bulb picked is defective," and let B,, Bs, and By denote that the bulb came from factory 1, or I, oF IIL, respectively. We wish to caleulate P(E), Now, the given data can be expressed in terins of probabilities. ~ Since 25% ofthe bulbs come from factory I, we have P(B,) =.25. Similarly, P(B,) = 35 and P(B,) =. : Since 3% of the bulbs from’ factory 1 are defective, that means that P(B| B,) = 03. Likewise, P(E | B,) = 05 and P(E | By) = 04 Substituting these six known probabilities into the right side ofthe equation of Theorem 7.2, we get P(E) = P(B,)PE | B,) + P(B:)P(E | B,) + P(B,)PLE | By) .25)(.03) + (.35)(05) + (40)(04) = 041 ‘Thus, the probability is.041 [or 4.1%] that a bulb picked at random isdefect B. Bayes’ Formula The terminology and assumptions used in the development of Bayes’ Formula are the same as those introduced in the discussion pertaining to Theorem 7.2. Let B, denote any of the quantities B,, B3, ..., By. Suppose that the Quantities P(B,), P(B,)...., P(B). and PCE | By), P(E | B,),..., PCE | B) are all known or can be determined easily, It is often important in practical situations to be able to calculate P(B;| B), the conditional probability of B, given E. The following equation, known as Bayes’ Formula, expresses P(B, | E)in terms of P(B,), P(E | B), and PE). ‘Theorem 7.3 (Bayes’ Formula): PBB oe B) P(E), is given in Theorem 7.2 ‘An equation for the denominator ten 163 Probability and Statistics EXAMPLE 7-13: Prove Bayes’ Formula. Proof: From the definition for conditional probability [Definition 3.1], we have PEOB) PE) (BIE) oO) We can replace the numerator term by P(B,n E) since EM B, = BOE, ‘Then, we apply the multiplication rule [Theorem 3.1] to P(B;o B) as follows: PAB. E) = P(B)PCE | B) @ Replacing the numerator of Bg, (I) by the right side of Eq, (2) results in Bayes’ Formula as given by Theorem 7.3 above. o Example 7-14 illustrates the usefulness of Bayes’ Formula, EXAMPLE 7-14: Refer to Example 7-12, which deals with light bulbs manu- factured at three factories of the GI Company. Suppose, we have picked out a defective light bulb from the batch of light bulbs discussed in Example 7-12. What is the probability that the light bulb came (a) from factory I, b) from factory I, (© from factory IIT? Solution: Let us continue to use the symbols introduced in Example 7-12. (a) Here, we wish to calculate the conditional probability P(B, |). Observe that P(B,) = 25, P(E | By) =.03, and P(E) = 041 from Example 7-12. Substitu- tion ofthese values into Bayes’ Formula yields P(B,)PCE | By) _ (251.03 PE) Oi ‘Thus, the probability is.183 [or 18.3%] thatthe light bulb came from factory I given that it is defective. (©) Proceeding as in part (a), we have PUB)PCE| Bs) _ (35)(05) PI) oar (6) Again proceeding as in part (a), we have PUB)PLE| Bs) _ (40604) PE) oar PB, |E) 183 PBI) 427 P(By| E) = ‘Observe that the sum of the probabilities from parts (a), (b), and (c) in Example 7-13 is 1. This is no accident, but is a confirmation of Theorem 7.4, which follows. The symbols that occur in Theorem 7.4 are those described and employed in Theorems 7.2 and 7.3. Theorem 7.4: Suppose that sample space S is partitioned into k events By, By, «.., By, and that E is any event of sample space S. Then P(B, | B) + PCB, | E) +... + PUI EY = 1 notes (a) A special case of Theorem 7.4 ovcurs in Theorem 3.4, which states that PB | A) = 1 — P(B| A). In this equation, B, BY, and A replace By, Ba, and E, respectively, of Theorem 7.4. Here, a partition of § is formed by Band BY, and k Discrete Probability Problems and Bayes' Formula 169 (b) Another special ease arises when Eis S, the sample space. In such a case, P(B,| E) = P(B,| S) = P(B),; and Theorem 7.4 reduces to PCB) + PB) +... + PB) =I o Equation (I) provides us with a consistency check since the collection of BB's constitutes a partition of S. [Thal is, we have S = B, U By U...U Bey where the B's are mutually exclusive.] AMPLE 7-15: Prove Theorem 7.4, Proof: Here, we have a pactition of $ into k events By, By, ...5 By. That is, $= BU B,U-..U By where BB, = @ whenever i # j Since the diagram for the k = 3 case is given in Figure 7-3, let’s do the proof for that case, First of all, we have HEB) MBO ings Oy PUB iy BB Thus, (B29) + PUB, OE) : PIB) From Figute 7-3, we see that £ = (By 6 E) (By 0 E)U (By 7B), where the three events on the sight of the form (B, > E) are matually exclusive [disjoint Thus, from Axiom 1.3', we have. PUB, | E) + PCB, |B) + PCBs | E) = ® P(E) = PB, B) + PUB, E) + PUBS) @ Substituting P(E) from Eg. Q) ito the denominator of Eq, (2) yields PUB, | E)-+ PCB) | E) + P(Bs | E) = 1 aa SOLVED. PROBLEMS Elementary Probability Applications PROBLEM 7-1 Carla and Dave each toss a coin twice. The one who tosses the greater number cof heads wins a prize. Suppose that Dave has a fair coin [P9(F1) = .S), while Carla has a coin for which the probability of head on a single toss is 4 [P:(H) = 4]. What is the probability that (a) Carla will win the prize (b) a tie results, () Dave will win the prize? Solution Preliminaries: Let C, stand for Carla's getting exactly one head, and similarly for the symbols Co and C,. Let D; stand for Dave's getting exactly two heads, and similarly for the symbols Dz and D;, Now, let us determine the individual probabilities for C, and D, for i = 0, 1, and 2. For example, P(C,) = PHT) + Pe(TH) = 2(4)(6) = 48, and P(D,) = P,(HH) = (3)? = 25. All the P(C,) and P(D)) values are presented in Table 7-3, which follows. Observe that the P(C)) values are binomial probabilities for n = 2.and p= .4, while the P(D,) values are binomial probabilities for m= 2and p = 5. [See Theorem 5.5,] @ (Carla wins) = P[(C, and Da) or (C, and D,) or (C, and D,)] x = P(C, atid Do) + P(C; and Dg) + P(C; and Dy) = P(C,)P(D,) + P(C:)P(Do) + P(C:)P(D,) = (48) 294 1125) + (1691.50) = 24 170 Probabitiy and Statistics o PiTic) PL(Co and Dg) of (C; and D,) or (C, and D;)] = P(Cy and Dg) + P(C, and D,) + P(C; and D,) = P(Co)P(De) + P(C,)P(D:) + P(C)P(D2) = (36)(.25) + (48)(.50) + (16)(.28) = 37 © P(Dave wins) = P[(D, and Co) or (Dz and Cy) or (D; and C,)} = P(D, and Cy) + P(D, and C) + P(D, and C,) = P(D,)P(Co) + P(D3)P(Co) + PUD2)PC,) = (.50)(36) + (25)(36) + (25)(48) = 39 An alternate approach for P(Dave wins) is the following: P(Dave wins) = | — [P(Carla wins) + P(Tie)} = 1 ~ [24 + 37] = 39 ‘TABLE 7-& Probabilities when Carla [C] and Dave [D] Fach Toss a Coin ] ‘Twice [Problem 7-1] | PC) PD) ga 36 | S= 25 2G = 48 | 257 = 50 #= 6 | t= 25 Sum 100 100 PROBLEM 7-2. For 2 machine consisting of four separate components, the probability of successful operation for each component is.95. What isthe probability the machine will work (a) if the machine will work only if all the components operate successfully (b) ifthe machine will work when any three of the components operate successfully? ‘Assume for both parts (a) and (6) that the components funetion independeily of each other. Solution Preliminaries: Let C, for i= 1, 2, 3, 4 stand fot “component i operates successfully.” Thus, P(C,) = PIC;) = P(C3) = P(Cx) = 95. | @ P(mnachine will work) = P(C, and C, and C, and C,) = P(C,C,C3C4) = PC)P(CL)P(CS) PICA) = (95}* = 8145 o | (b) Let G; indicate that component ! doesn't operate successfully, for f= 1, 2, 3, 4 Since the machine will work when any three of its components operate successfully, there are four now orders for which the machine will work. These are as follows: C,C,CsCi, C,C,C\Cas QCiCsC, CyCsC,Cy. Now, PIC) = 1 — P(C) =.05. The probability for C,C,CsCi. is calculated as follows PUCACACQCA) = PIC)PUC2)PLCS)PCC ‘The probability isalso(95)"(05)for each of C, Cy we have (machine will work (95)%405) @ °3C3Cyand Cj C,C,C,. Thus, PUCLCACyC4) # 4: PLCC C3 Ca) = (95)* + 4. 95)5(05) 8145 + 17148 = 9860 ® PROBLEM 7-3 (a) Given the integers 1, 2, 3,..., nm. Develop an equation for the probability that a randomly chosen permutation of these integers is a derangement. [Recall that, in such a permutation, no integer appears in its natural place. For a review of these and related ideas, refer to Examples 7-8, 7-9, and the discussion following Example 7-9.] (b) Suppose that seven chips, numbered 1, 2,3, 4, 5, 6, 7 are randomly drawn out one at a time, from a container, and placed in envelopes labeled, respectively, with the numbers 1, 2,3, 4, 5, 6, 7. [The first chip drawn out Discrete Probability Problems and Bayes’ Formula 171 ‘goes to envelope 1, the second chip to envelope 2, etc.] What is the probability that every envelope ‘will contain a chip whose number is different from the number labeled on the envelope? Sotution (a) In Example 7-9, we stated the equation for P(A, UA,U...UA,) which is equal to the probability that a randomly chosen permutation of I, 2, 3 ...,m has at least one integer in its : natural place. Now, the complement of the event 4, UA, U...U A, isan event that consists i ofall the derangements of the integers 1, 2, 3,...,. Thus, = PAU ARU AGUA) tg curd] yt teas wo {derangement of 1,2,...,») {Note that the (— 1)" in the last term arises from —(—1¥"*.] The limit ofthis probability as ‘n becomes infinite is e~, which equals 36787944 to eight significant digits. The probability expression of Eq. (I) agrees with the value of e~* to at least six significant digits for n = 9 In this regard, see Table 7-2, which follows Example 7-9. Table 7-2 presents values of P(A, 0A, U...W Ay) as a function ofn for m between 2 and 10, inclusive. (©) The probability here is given by the Eq. (1) result for n = 7. Thus, the probability that every ' envelope will contain a chip with a number diferent than the number on the envelape is given by tit atah 367857 hl Total Probabil ty Theorem and Bayes’ Formula PROBLEM 7-4 An appliance store receives shipments ofa certain type of radio from three of| its warehouses: 2074 come from warehouse I, $0% from warehouse Il, and 30% from warehouse EL It is known from past experience that 5% ofthe radios from warehouse I are defective, while the corresponding percentages of defectives are 4%, and 7% for warehouses Il and II] respectively. (a) If single radio is picked at random from the inventory atthe store, what is the probabi is defective? [Equivalently, what isthe fraction of defective radios among all the radios shipped i to the appliance store from the three warchouses?] (b) What isthe probability that a radio came | from warehouse I if (equivalently, given that] it turns out to be delective? Solution Preliminaries: Let D stand for a “defective radio,” and let R, indicate that a radio comes iM from warehouse I, with similar interpretations for R, and Ry (a) From the theorem of total probability [Theorem 7.2] and the given data, we have P(D) = P(R,)P(D| R,) + P(R3)P(D | Rs) + P(R3)P(D | Rs) = (2)(05) + (5)(04) + (3)(07) = 051 ‘Thus, the probability of picking a defective ratio is 051 or 5.1%. Equivalently, the fraction of defective radios among all the radios received by the appliance store is 051 (b) Here, we want (0 calculate P(R, | D). From Bayes’ Formula [Theorem 7.3] and the given data, we have POR,)P(E| Rs) _(2)(05) PAR 2 ee P(E) 051 ioe Here, the denominator value is from part (a). Thus, the probability is .196 that a radio came from warehouse I, given that it is defective. PROBLEM 7-5 A new test for detecting cancer has been developed. Suppose that 90% of the i ‘cancer patients ina large hospital reacted positively to the test, while 15% of the remaining patients ij 172 Probability and Statistics in the hospital reacted positively to the test. [Assume that none of these remaining patients has | cancer. That is, assume that the only patients in the hospital who have cancer are the cancer patients. Also, observe that a positive reaction to the test means that the test indicates that a person has cancer. That is, itis possible for a person who does not have cancer to show a positive reaction to the test. This is known as a false positive.) Suppose that 4% of the patients in the hospital are cancer patients [equivalently, 4% ofthe patients in the hospital actually have cancer], ‘What isthe probability that a hospital patient picked at random who reacts positively to the test is a cancer patient? | Solution Let C and C’ denote, respectively, the events that a patient is a cancer patient, or is not ‘a cancer patient [i.c, is a "remaining patient"). Let Pos denote that a patient reacts positively to the test. We wish to caleulate P(C | Pos). From Bayes’ Formula [Theorem 7.3], we have P(Pos) = P(C)P(Pos | C) + P(C’)P(Pos | C’) = (04)(.90) + (96)(15)=.18 (2) a 04) 32m which means that there is a 20% probability that a patient who reacts positively to the test is a cancer patient. PC | Pos) = a P(C | Pos) = @ PROBLEM 7-6 This problem deals with the subject of Bayesian decision theory, which is based, in part, on Bayes’ Formula, Having commissioned a survey of oil deposits on adjoining regions, a landowner estimates that the probability of an oil discovery on his land is 2. Let S, and 5, indicate, respectively, the presence or the absence of oil on his land. Thus, P(S)=2 and POS) 40, Q) ‘These are known as a priori or prior probabilities. w ‘The landowner decides to have expensive sonar testing [involving sound waves] done on his "| land. Let 6, designate the event that the sonar testing indicates the presence of oil, and 8, designate theevent that the sonar testing indicates that no oil is present. The engincer doing the sonartesting | admits that 10% of the time the test will indicate no oil when, in fact, oil is present. That is, PQz|S))=1, andhence —P(6,|S,)=.9 @.@) | Also, when oil is not present, the test indicates no oil 607, of the time. That is, | P(O;|S,)=.6, andhence —P(0, |S) = 4 6) [The results derived in Eqs. (4) and (6) follow directly from Eqs. (3) and (S), respectively, as a consequence of applying Theorem 7.4 for the case where k = 2.] Find P(S; |@,) and P(S; |€:) which are the a posterior or posterior probabilities that oil is present, given, respectively, postive and negative results of sonar testing Solution First, we find the [total] probability P(9,) that sonar testing will show the presence of oil, From the total probability theorem [Theorem 7.2], or from common sense, POs) = P(S,)PCO, |S.) + PUS2)POs |S) = (29) + (84) = 50 ‘Thus, by Bayes’ Formula [Theorem 7.3}, PES) P(@ 1S) _ (29) _ POS, 180)= By Ty 36 ®) and thus, POS, |, — PIS, 1) = 64 O) Discrete Probability Problems and Bayes’ Formula Now, it should be clear that P(0,) using Theorem 7.2, we have — P(0,), and hence that P(0;} = 50. Asa check, after P(Ba) = P(S,)P(G, | S,) + P(S2)P(G2 | So) = (2)(1) + (8)(.6) = 50 (10) ‘Thus, by Bayes’ Formula, P(S,)PAO, (209 PAS, 162) = St = au) and thus, PUS,| 8) = 1 ~ PUS; |) = 96 cr) Note that the posterior probability of finding oil given a positive test result, namely, .36, is greater than the prior probability of finding oil, namely .2. The posterior probability of finding oil given a negative test result, namely .04, is lower than the prior probability of finding oil. Here, ‘we havea realisticexample of subjective probabilities being influenced by the appearance of further information, Supplementary Problems PROBLEM 7-7 At Milton College tis known that 75% of the students who took three years of high school mathematics will pass.the Elementary Statistics course, while 45% of those who didn't take three years of high schoo! mathematics will pass the Elementary Statistics course. [These figures apply to the-first attempt at taking the Elementary Statistics course. Ifa student has not passed the course that means that the student has either failed the course or has not completed the requirements for the course.| I is also known that 40% of the students at Milton College who attempt to take the Elementary Statisties course have taken three years of high school mathematics. A student is picked at random from among those students at Milton College who have already attempted to take the Elementary Statistics course. (a) What is the probability that the student has passed the Elementary Statistics course on the first attempt? (b) What is the probability that a student has taken three years of high school mathematics given that the student has passed the Elementary Statistics course on the first attempt? Answer (a) S10. (b) 30/57 526 PROBLEM 7-8 Refer to Example 7-2. What is the probability that testing is required of (a) exactly two bolts, (b) exactly three Bolts, and (@) exactly five bolts? Answer is (WO note: Suppose that random variable X deriotes the exact number of bolts that are required to be tested, Then, the probability function for 4 is given as follows: f(2) = 3, (3) = (4) = as [rom the solution to Example 7-2}, and f(5) = fi PROBLEM 7-9. Ofall the job applicants at the Apex Company who come for a job interview, 30% are given the interview right away [say, within $ minutes of arriving]. The other 70% are asked to wait-in.a.waiting room. About 40% of the time, people who are asked to wait will leave the waiting room before being given the job interview, and will not return. (a) What percent of the job applicants who come for a job interview will actually have the interview? (b) Given that aan applicant is given an interview, what is the probability that the applicant had to wait in the waiting room? Answer (a) 7%, (W142! = 58S 3 174 Probability and Statistics PROBLEM 7-10 Refer to Example 7-5. For a group of three people chosen randomly from the ‘overall population, what is the probability that (a) exactly two wete born in the same quarter of the year and (b) all throe were born in the same quarter of the year? Answer (a) 36/64 = 9/16 = 5625 (b) 4/64 = 1/16 ~ 0625 PROBLEM 7-11 Each of three boxes contains two balls. One box contains two green balls, ‘another contains a green and a red ball, and the third contains two red balls. A box is chosen at random, and one ball is taken out, What is the probability that the second ball in the box is green, given that the ball that is taken out is green? Answer 23 PROBLEM 7-12 Box I contains two green and three red balls, box II contains four green and two red balls, and box II contains three green and three red balls. Suppose one ball is drawn | from each box. What is the probability that a green ball was drawn from box I, given that two green balls were drawn? Answer 2 PROBLEM 7-13. The Slobovian army believes that theirencmy, the Grimalkin army, will attack at locations 1, 2, and 3 with probabilities 4, 25, and 35, respectively. [Note that the probabilities ‘add up to 1.0} In symbols, P(A,) = 4, P(A3) = 28, and P(A) = 35. ‘A Slobovian reconnaissance team then observes the Grimalkin army setting up long range cannons. Let this action be designated as action B. The Slobovian intelligence experts believe that the probabilities of action B, given a plan to attack at locations 1, 2, and 3, respectively, are land 3. Use Bayes’ Formila [Theorem 7.3] to determine posterior probabilities thatthe Gri army will attack at locations 1,2, and 3, respectively. Answer P(A |B) = 258, P(As | B) = 403, PCAs |B) = 339 PROBLEM 7-14 Suppose Bertha holds two tickets from a lottery in which 1000 tickets were sold. The tickets have been numbered from 1 to 1000. Ten prizes are to be awarded. [That is, ten of the 1000 tickets will be designated as prize-winning tickets.] Let X be the number of prizes that Bertha will win, Determine the probability function of X. ; oj t|2 : 98009 | 01982 ! 0008 Answer Fea | PROBLEM 7-15 Refer to Problem 7-1 again. In the experiment of that problem, Dave tosses a fair coin twice while Carla also tosses a coin twice, For Carla's coin, the probability of head on ‘single tossis 4. What is the probability that Dave will win the prize provided that the experiment is repeated whenever a tie occurs? [Suggestion: Make use of Theorem 6.5, which gives the equation for the sum of a geometric series.] Answer 13/21 = 61905 i i f MIDTERM EXAM (Chapters 1-7) 2S ST A PE TE 1. Given a container with three red balls, labeled 1, 2, 3, and two green balls, labeled 4 and 5, Suppose two balls are dravmn in succession, resulting in three balls remaining in the container five, we have draving without replacement here}. Describe a sample space if(a) one wishes to record the number on each ball, if) one wishes to record the color on each ball, or if (c) ‘one wishes to record the total number of ted balls obtained, 2. For the sample space $; of Problem { above [see solution to part (a), indicate the following, events as subsets of S,: (a) Obtaining an odd number on both draws. [Call this event .] (b) Obtaining a sum of numbers equal to six. [Call this event B.] (e) Obtaining exactly one red ball. [Call this event C] 3, Determine the probabilities of the events 4, B, and C of Problem 2, Note that itis reasonable to assume that all 20 elements of sample space S, are equally likely 4. For people who visit Eddie's Fast Food Restaurant, 40% order a meal containing a hemburge, 35% order a meal containing a milkshake, and 10% order a meal containing both a hamburger and a milkshake. What isthe probability that @ petson visiting the restaurant will (a) have cither a hamburger or # milkshake or both, (b) will have exactly one of a hamburger or a milkshake, and (¢) will have nether a hamburger nor a milkshake? (2) What are fair odds in favor of the event of part (a)? 5. For customers who come toa restaurant, 45% are seated immediately. Of those who are asked to wait, 40% become impatient and leave the restaurant before being seated. (a) What percent ofall customers who come to the restaurant are actualy seated? (b) For those customers who are seated, what is the probability that they had previously been asked to wait? 6. A person is dealt five cards from an ordinary deck of 32 cards. What is the probability of being dealt a hand with (a) four spades and one heart, (b) exactly four spades, (e) exactly four cards of the same suit, and (@) at least four spades? 7. A pair of fair.dice are tossed. Let 4 be the event “even number on both tosses,” and let B be the event "sum of the values on the dice equals six.” (a) Display A as a subset, and compute P(A).(b) Display Bas a subset, and compute PB), (c) Compute P(B | A). (d) Compute P(A | Bp (6) Determine if A and B are independent events 8. Angie and Bobbie play for the handball championship. The one who wins two games first is the champion. [Thisis referred to as.a “best two out of three” series of games.] The probability that Angie will beat Bobbie in an individual game is 3/5. What is the probability that Angie will win the championship? 9. Givena container with six red and four white balls. Three balls are drawn without replacement Let X denote the number of red balls drawn, (a) Determine the table of values for f(x), the probability function for X. (b) Determine values of the cumulative distribution function F(x) for all x values. 10, In an office, 25%, of the 200 employees were smokers. A set of five employees is randomly selected. (a) What is the probability that exactly three of them are smokers? (b) Use the binomial distribution to approximate the answer to part (3) HL, A large shipment [several thousand] of radios contains 2.5% that are defective. Suppose that 100 of the radios are randomly selected. What is the probability that (a) exactly three will be defective, (b) at most three will be defective, and (e) at least four will be defective? (€) Rework parts (a), (b), and (c) by using a Poisson approximation to the binomial distribution. vs 176 Probability and Statistics 12, Ina container, 40% of the chips are red and the rest are of other assorted colors. Chips are drawn from the container, with replacement after each draw. (a) What is the probability that ‘occurs on the eighth chip drawn? (b) What is the probability that the first red chip occurs on or before the eighth chip drawn? (e) What is the probability that the third red chip occurs on the eighth chip drawn? 13, Carriers of a certain disease constitute 2.5% ofa certain population. There exists a blood test for which the probability is.98 that a cartier of the disease will test positive, and for which the probability is 99 that a noncarrier will test negative. (a) For a person selected at random from the population, what is the probability that the person will test positive? (b) Ifa person tests positive, what isthe probability thatthe person is a carrier ofthe disease? 14, Twro defective batteries have been mixed in with four good ones. Suppose that batteries are tested one by one, until both defective batteries are found. (a) What is the probability that the testing of exactly five batteries is required? (b) Let random variable X denote the exact number of batteries to test so as to find both defective batteries. Determine the pf (probability function] for X. Solutions to Midterm Exam 1. (a) Here, for example, element 25 indicates two on the first ball and five onthe second ball. Thus, using the label S, for the sample space, we have 5, = (12,13, 14, 15,21, 28,24, 25,31, 32, 34,35, 41,42, 43,45, 51, 52,53. 54} [Note that §;, which contains 20 clements, i the most subdivided sample space for the experiment of drawing two balls without replacement. ie () Here, S; = {RG, RR,GG, GR}, ithe ordering is of interest, and S, = (rg,rr,ggh if ordering isnot of interest. Here, the element rg corresponds to RG and GR of S;, o) Here, 5, = (0,1,2}, where the elements refer to the numberof red balls obtained. 2A {13,15,31,35,51,53)_ () B= (15,24,42,51) cat 4, 15,24, 25,34, 35,41, 42,43, 51, 52,53} 3, Ftom Theorem 1.2 [Section 1-SB], we know that ifll the elements ofa sample space S are equally likely, then P(E) = N(EVN(S) for any event [subset] & of S. Thus, sinee N(S;) = 20, P(A) = N(AYN(S)) = 620 30 P(B) = N(BYNIS,) = 4720 20 and PIC) = N(CYNIS,) = 12/20 = 60 4, Preliminaries: Refer tothe Venn diagram of Figure M-I, where H refers to “hambur- gers," and M refers to “milkshakes.” For the disjoint regions, we have Ry = HM, R, ~ HOM, etc. Thus, P(R,) = -10, P(R,) = 40 ~ 10 = 30, P(R,) = .35—.10 25,and P(R,) = 1 (10-4 30+ 25) = 35, (a) This event is HUM, and P(H UM) = PUD + POM) ~ PU OM) PUR,) + PRs) = 65. (0) In terms of disoint regions, this event is Ra Rs. Now, PRs Rs) = P(Ra) + PAR) = 55. (6) In terms of disjoint regions, this event is Ry, Thus, Rad PCoeither hamburger nor milkshake") = P(R,) ~ 35 Figure M-1. Venn diagram for (@) Fair odds for the event Hw M are equal tothe ratio of PCH UM) to PLU MY, Problem 4. that i, of 65 10.35, or of 65 t0 35, oF of 13 107 Midterm Exam (Chapters t-7) 177 ‘Thos, substituting from Eqs. (8) and (10) into Eq. 8), ovat | seated) = 33/678) = 423, ‘5, Preliminares: Ref to Figure M-2. From Figuse M-2, we see that Seen) _ Aid oii) B=CUD whore Cad=Qlie,Cand Daredisioint) a) | a4 Waianae scat a) Prrseated”) = PAU) = PLA)+ PI) ® (3) Now, C= BrsC since C & B, and thus, ee s PO) = PBC) = PBC |B) % eee Bu, PB) = PLA)= 55 and (4) Figure M-2. Venn diagram tor Reinet cms e 3 Pablms- By8 wien Gen ‘Thus, substituting Bas. (4) and () in Ba.) i PIC) = (55)(60) = 33 G) Substituting the values for P(C) and P(A) into Eq, (2) we get | Peseated") = 45 + 33 =.78 o | (0) First ofall, : ‘Pat | seated) = P(wait seated Pscated) ® i From path t Plsemed) = PAL) = © i Also, from pat (a), I Plwait m seated) » P(C) = 33 19) i 6. In all parts ofthis problem, we have drawing without replacement of five cards oben n= 2)(2) (2) worsen = ass 5 (b) (exactly 4 spades) = P(4 spades and 1 nonspade} (13) /39\ {52 =(3)0Y 3) = (dy P(at least 4 spades) = P(exactly 4 spades) + P(exactly 5 spades) -[()€)-@)UG) =») =m P(A) = Pleven, and ovens) = (P{even) = ( a (0) Here, B = (15,24,33,42, 51), and P(D) = NBYNIS) = 576, (©) Since dB = (24,42), Plain B) = 2/36, hence P(B| A) = PLA BYPCA) = 29 @) PAL B) = Plan ByPLB) = 2/5 fe) Since P(A B) = 2/36 = 0555, and P(A)P(B) = (21%) independent 03472, events A and B are not Ee & Let A, stand for the event that Angie wins game i, for U2 1, 2, 3, with a similar meaning. f for B, with respect to Bobbie. Thus, 178 Probability and Staistics (Angie wins champ) = P(A, Ay) + P(A Bp) + P(BALAy) wo We assume the individual games are equivalent to independent trials, Thus, P(A, Bad) = PlAy)P(B,)P(As) = GPG) = Hs ® and similatly for P(A, A,) and P(B, A, 43) Thus, (Angie wins champ) = (3) + 207°) @ 9, (a) Here, we have a hypergeometric pf given by “fe8) = Alas, M) = i 3, 6,10) = Gen) ait ® ~ 1 ‘Thus, we generate the following table of values of fx}. x] o | Tei |ea 3 [sum Fe) | 4j120 | 36/120 | eojr20 | 2020 Ft () The table for the ed, F(x), is as follows: interval | x<0 | 0sx

    You might also like