DMD课件及资料 chapter2_第1页
DMD课件及资料 chapter2_第2页
DMD课件及资料 chapter2_第3页
DMD课件及资料 chapter2_第4页
DMD课件及资料 chapter2_第5页
已阅读5页,还剩112页未读 继续免费阅读

下载本文档

版权说明:本文档由用户提供并上传,收益归属内容提供方,若内容存在侵权,请进行举报或认领

文档简介

1、1,Chapter 2 Fundamental of Discrete Probability,Data, Models, and Decisions,2,Milestones in Probability Theory,1654: B. Pascal and P. Fermat (French) derived the exact probabilities for certain gambling problem. 1657: C. Huygens published the first book on probability.(Holland) -论赌博中的计算 1713: J. Ber

2、noulli published the book, The Art of Conjecture. (Switzerland) 第一个极限定理-伯努利大数定理:在多次重复试验中,频率有越趋稳定的趋势 猜度术 1882: Pierre de Laplace introduced many mathematical techniques. (French) 拉普拉斯-严密的、系统的科学概率论的最卓越的创建者,概率的分析理论 1931: R. von Mises initialized the concept of sample space, and filled the gape between

3、probability theory and measure theory. (Austria) 奥地利数学家冯.米西斯提出的公理理论并不完善 1933: A. Kolmogoroff gave a complete axiomatic foundation of probability theory in his book, Foundations of Probability Theory. (Russia)苏联数学家柯尔莫哥洛夫用公理化结构明确定义了概率,成为概率论发展史上的一个里程碑,3,法国数学家帕斯卡与费马往来的信函中讨论合理分配赌注问题。该问题可以简化为: 甲、乙两人同掷一枚硬币

4、。规定:正面朝上,甲得一点;若反面朝上,乙得一点,先积满3点者赢取全部赌注。假定在甲得2点、乙得1点时,赌局由于某种原因中止了,问应该怎样分配赌注才算公平合理。 帕斯卡:若再掷一次,甲胜,甲获全部赌注, 两种情况可能性相同,所以这两种情况平均一下,若乙胜,甲、乙平分赌注。 甲应得赌金的3/4=1/2+(1/2)*(1/2),乙得赌金的1/4=(1/2)*(1/2) 。 费马:结束赌局至多还要2局,结果为四种等可能情况: 情况 胜者 甲甲 甲乙 乙甲 乙乙 前3种情况,甲获全部赌金,仅第四种情况,乙获全部赌注。所以甲分得赌金的3/4,乙得赌金的1/4。,4,Chapter Outline,2.1

5、 Outcomes, Probability and Events 2.2 The Laws of Probability 2.3 Working with Probability and Probability Tables 2.4 Random Variables 2.5 Discrete Probability Distribution 2.6 The Binomial Distribution 2.7 Summary Measures of Probability Distribution 2.8 Linear Functions of a Random Variable 2.9 Co

6、variance and Correlation 2.10 Joint Probability Distributions and Independence 2.11 Sums of Two Random Variables 2.12* Some Advanced Methods in Probability,5,随机试验(Random Experiment) 基本事件(Outcome) 互斥性(Mutually Exclusive) 完备性(Collectively Exhaustive) 概率(Probability) 事件(Event),2.1 Outcomes, Probability

7、 and Events,Main Concepts,6,随机试验例一,条件:掷一枚硬币 观测特性:正、反面出现的情况 基本事件:正面H,反面T,7,随机试验例二,条件:将一枚硬币掷三次 观察特性:正、反面出现的情况 基本事件:HHH,HHT,HTH,THH, HTT,THT,TTH,TTT,8,随机试验例三,条件:将一枚硬币抛掷三次 观测特性:出现正面的次数 基本事件:0,1,2,3,9,随机试验例四,条件:抛一颗骰子 观测特性:出现的点数 基本事件:1, 2, 3,4,5,6,10,随机试验例五,条件:电话交换机呼唤次数 观测特性:一分钟内接到的呼唤数 基本事件:0,1, 2, 3, .,1

8、1,随机试验的特点,试验可以在相同条件下重复进行; 试验的全部可能结果不止一个,但试验之前能明确知道所有可能的结果; 每次试验必发生全部可能结果中的一个且仅发生一个,但某一次试验究竟发生哪一个可能结果在试验之前不能预言。,12,基本概念解释,随机试验:满足上述三个条件的试验叫随机试验,简称试验; 基本事件:试验的一个可能结果; 样本空间:全体基本事件构成的集; 事件:由一组基本事件组成的,事件是样本空间的子集;,13,基本概念解释,互斥性:是指任意两个基本事件不能同时发生; 完备性:如果一组事件两两互不相容,其和等于样本空间,则称这一组事件为一完备事件组,例如所有基本事件就组成一个完备事件组。

9、 下雨的概率为50%,怎么理解?,14,2.2概率的性质(Laws of probability),任何事件的概率0 p 1 性质一:任何事件的概率都是0和1之间的一个值。概率越大事件发生的可能性也越大。概率为1的事件必然发生;概率为0的事件不可能发生。 性质二:如果事件A和事件B互斥,则,15,2.2概率的性质(Laws of probability),条件概率(Conditional probability)P(A|B)表示在知道事件B已经发生的条件下,事件A的概率,性质三:如果A和B是两个事件,则,16,例2.1 纸牌游戏中的条件概率,事件A表示摸到的牌是Q ,事件B表示摸到的牌是一张人

10、像牌。事件A/B表示在我们摸到一张人像牌的条件下摸到的牌是Q,17,例2.2 班里随机选学生,100人的班级,40名男生,60名女生; 70名本地学生, 30名国际学生; 70名本地学生中有25名男生和45名女生; 30名国际学生中有15名男生和15名女生。,18,分别用A、I、M和W表示选出的学生是本地学生、国际学生、男生和女生的事件。 P(A)=0.70,P(I)= 0.30,P(M)=0.40,P(W)=0.60。 在所选择的学生是男生的情况下,该生是本地学生的概率P(A/M) 或=25/40=0.625,100人的班级,40名男生,60名女生; 70名本地学生, 30名国际学生; 70

11、名本地学生中有25名男生和45名女生;,30名国际学生中有15名男生和15名女生。,19,2.2概率的性质,性质四:如果事件A和B是独立事件,则 如果A和B是独立事件,则,20,两个基本概念,独立(Independent):事件A发生的概率并不影响事件B发生的概率,称事件A和事件B相互独立。(样本放回实验) 不独立(Dependent):事件A发生的概率影响事件B发生的概率,称事件A和事件B相互不独立。(利息降低和股指增长之间的关系, 样本不放回实验 ),21,概率计算的一般步骤,定义不同事件来表示问题的不确定性; 用公式表示出这些事件之间的关系,例如条件概率(“A|B”),与(“AB”),或

12、(“A+B”); 用表格表示所有相关事件的概率; 使用概率的性质计算概率,22,2.3 Working with Probability and Probability Tables(1),Example 2.3Using Probability Tables to Compute Probabilities in a Decision Tree,Development of A New Consumer Product Caroline Janes: marketing manager Product: “Suds-Away”- a new automatic dishwashing det

13、ergent Market: Strong (30% of chance), make $18 million; Weak, lose 8 million Market survey: cost $2.4 million; If market is weak, there is 10% chance that the test will be positive; if market is strong, there is 20% chance that the test will be negative.,Caroline can decide either to not produce Su

14、ds-Away, to conduct the market survey test prior to deciding whether or not to produce, or to go ahead with production without conducting such a market survey test.,23,A,B,C,E,G,F,D,Strong,Weak,0.30,0.70,$18 million,-$8 million,Produce,Do not produce,-$2.4 million,Strong,Weak,$15.6 million =18-2.4,-

15、$10.4 million =-(8+2.4),P3=?,P4=?,P5=?,P6=?,P1=?,P2=?,Do not produce,$0,No survey, produce,Conduct market survey test,Positive survey results,Negative survey results,Produce,Do not produce,-$2.4 million,-$10.4 million,$15.6 million,2.3 Working with Probability and Probability Tables(2),24,2.3 Workin

16、g with Probability and Probability Tables(3),S- Market is strong Q- Market test result is positive W- Market is weak N- Market test result is negative,We have known that: P(S)=0.30 P(W)=0.70 P(Q|W)=0.10 P(N|W)=0.90 P(N|S)=0.20 P(Q|S)=0.80,P(QW) = P(Q|W)P(W) = 0.100.70 = 0.07 P(NW) = P(N|W)P(W) = 0.9

17、00.70 = 0.63 P(QS) = P(Q|S)P(S) = 0.800.30 = 0.24 P(NS) = P(N|S)P(S) = 0.200.30 = 0.06,Market is strong Market is weak Total,Market test is positive Market test is negative Total,0.24 0.07 0.31 0.06 0.63 0.69 0.30 0.70 1.00,Complete probability table for Caroline Janes decision problem,P(Q),P(N),25,

18、A,B,C,E,G,F,D,Strong,Weak,0.30,0.70,$18 million,-$8 million,Produce,Do not produce,-$2.4 million,Strong,Weak,$15.6 million,-$10.4 million,P3=?,P4=?,P5=?,P6=?,P1=?,P2=?,Do not produce,$0,No survey, produce,Conduct market survey test,Positive survey results,Negative survey results,Produce,Do not produ

19、ce,-$2.4 million,-$10.4 million,$15.6 million,2.3 Working with Probability and Probability Tables(4),P1=P(Q)=0.31 P2=P(N)=0.69,26,2.3 Working with Probability and Probability Tables(5),9.73,-8.14,9.73,-2.4,1.36,-0.2,1.36,27,2.3 Working with Probability and Probability Tables,Example 2.4Quality Contr

20、ol in Manufacturing 制造业中的质量控制 A manufacturing process produces microprocessors using an entirely new technology. Historical results suggest that 70% of the final product is damaged and therefore not usable. In the past, the manufacturer has used an extremely expensive (but extremely reliably) diagno

21、stic procedure to test if a microprocessor is damaged. In order to cut costs, management is considering using a very inexpensive test that is not as reliable. Preliminary data suggest that if a microprocessor is damaged, then the probability that the new test would be negative is 0.80. If the microp

22、rocessor is working properly, the probability that the test is negative is 0.10.,28,2.3 Working with Probability and Probability Tables(7),Two types of errors with the new test: A damaged microprocessor might test positive A perfectly good microprocessor might test negative Management would like to

23、know the probabilities that each of the two errors might occur.,29,Set up: D the event that a microprocessor is damaged W the event that a microprocessor is working properly Q the event that a microprocessor tests positive N the event that a microprocessor tests negative We would like to know: P(D|Q

24、) and P(W|N),2.3 Working with Probability and Probability Tables,30,Having known that: P(D)=0.70, P(N|D)=0.80, P(N|W)=0.10 From these, we get: P(W)=0.30, P(Q|D)=0.20, P(Q|W)=0.90 (性质2) P(D and Q)= P(Q|D)* P(D)=0.14 (性质3) The probability table:,2.3 Working with Probability and Probability Tables,31,C

25、onclusion The test is relatively reliable when it is negative for only 5% of microprocessors that test negative would be working properly. But, there will be 34% of microprocessors that test positive (and therefore would be shipped to customers) would be damaged.,2.3 Working with Probability and Pro

26、bability Tables(10),what is your recommendation to the management based on the above results?,Not employ this test,32,例2.5 一种新型客户服务方式成功的概率,一种新型客户服务方式成功的含义为该服务的需求大或竞争反应不迅速; 假设需求大的概率是0.60,则需求小的概率是0.40; 假设竞争反应迅速的概率是0.70,则竞争反应不迅速的概率是0.30; 假设在需求大的条件下竞争反应迅速的概率是0.90 。 这种新型客户服务方式将会成功的概率?,33,建立一个概率模型,事件H表示需求

27、大; 事件L表示需求小; 事件Q表示竞争反应迅速; 事件R表示竞争反应不迅速; 事件S表示新型客户服务方式成功。 我们需要计算P(S) ; P(S)=P(H或R),34,已知:P(H)=0.60, P(L)=0.40, P(Q)=0.70, P(R)=0.30, P(Q/H)=0.90. 于是,P(HR)=P(H)P(R/H)=P(H)(1-P(Q/H)=0.60*(1-0.90)=0.06. 所以,P(S)=P(H或R)=P(H)+P(R)-P(HR)=0.60+0.30-0.06=0.84.,Answers,这种新型客户服务方式将会成功的概率为0.84。,35,例2.6 谋杀案审判中的DN

28、A检验,Jacob是谋杀案审判中的陪审员。假设过去几天的审判中,他估计被告有罪的概率为p,p介于0.500.60。 DNA专家检验在案发现场发现的血迹跟被告的血型相符。 DNA专家使用的新的DNA测试方法是99%可靠的。 有了DNA检验结果之后,Jacob认为被告有罪的概率是多少呢?,36,DNA测试方法99%可靠的含义,如果血型属于同一个人,则DNA检验结果为匹配的可能性是99%; 如果血型不属于同一个人,则DNA检验结果为不匹配的可能性也是99%。,37,概率模型,事件G表示被告有罪; 事件I表示被告无罪; 事件M表示现场的血型与被告的血型相匹配; 事件N表示现场的血型与被告的血型不相匹配

29、; 事件D表示DNA检验的结果为匹配; 事件E表示DNA检验的结果为不匹配;,38,进一步假设,假设事件G和事件M是相同的(如果被告人有罪,那么血型相符;如果血型相符,则被告人有罪); p表示Jacob以前对被告有罪估计的概率; 我们的目标是计算出P(G|D)或P(M|D) 。,39,已知: ,P(N)=1-P(M)=1-p 由99%可靠得:P(D/M)=0.99, P(E/N)=0.99. P(E/M)=0.01, P(D/N)=0.01. P(DM)=P(D/M)*P(M)=0.99p, P(DN)=P(D/N)*P(N)=0.01*(1-p). P(EM)=0.01p, P(EN)=0.

30、99(1-p).,P(D)=P(DM)+P(DN) =0.99p+0.01(1-p)=0.01+0.98p. P(E)=P(EM)+P(EN)=0.01p+0.99(1-p) =0.99-0.98p.,40,当p=0.5时,当p=0.1时,结果显示了DNA检验的权威性,当DNA的结果是匹配时,Jacob估计有罪的概率就从10%上升到92%.,41,例2.7 制造系统的可靠性,两台机器M1和M2组成的制造系统; 如果M1和M2都正常时则这个系统运作正常; 如果有一天M1或M2中的一个不能工作则这个系统当天不能工作; M1和M2在给定的一天不能工作的概率都为 p=0.10,问题: 购买一台新机器M

31、3,在M1或M2不能工作的时候代替它们; M3在给定的一天不能工作的概率也为 p=0.10; 计算系统在给定的某一天系统可以运作正常的概率,无论是否购买了M3。,42,概率模型,事件Wi表示机器Mi运作正常,I=1,2,3; 事件Fi表示机器Mi不能工作, I=1,2,3; 事件S表示系统可以运作正常; 计算P(S) 。,43,已知条件和计算过程,已知: P(Fi)=p=0.10, P(Wi)=1-p=1-0.10=0.90,I=1,2,3。 没有M3的时: P(S)=P (W1 W2)= P (W1 )*P(W2) =(1-p)*(1-p)=0.9*0.9=0.81.,44,买了M3后能正常

32、工作的情况,M1和M2都正常运作; 或 M1不能正常运作,但M2和M3都运作正常; 或 M2不能正常运作,但M1和M3都运作正常。 所以,P(S)=P (W1 W2)或(F1W2 W3)或(W1F2 W3)。,45,加上M3后的计算过程,P(W1 W2)=(1-p)*(1-p) P(F1W2 W3)=p*(1-p)*(1-p) P(W1F2 W3)=(1-p)*p*(1-p) 当p=0.1时得, P(S)=0.972. 所以,增加新机器M3可以使系统的可 靠性从0.81增加到0.972.,46,概率模型计算的一般步骤:,定义不同的事件来表示问题中的不确定性; 用公式表示出这些事件之间的关系,例

33、如条件概率(“A|B”),与(“AB”),或(“A+B”); 必要时,用表格表示所有相关事件的概率; 使用概率的性质计算概率,可以提高对各种不确定性问题的认识和理解,47,2.4 Random Variables,前面介绍的概率模型或试验, 列出了某一不确定量的所有结果,以及相应的概率。 When the outcomes in a probability model are numbers, we will refer to this uncertain quantity as a random variables. 随机变量是定义在样本空间S=上的一个单值实函数,记作X=X(),简记为X

34、Two types of random variables: discrete r.v. assumes only distinct and separate numbers continuous r.v. takes any value within some interval of numbers,48,随机试验与随机变量,样本空间:,X表示射击中靶的次数,为随机变量,对应的取值为:0,1,2。 某公司第二年新型飞机订单数离散随机变量 某一天某饭店的就餐人数离散随机变量 某地某一阶段的降雨量连续随机变量,49,How to describe a random variable?,Proba

35、bility Distribution,50,2.5 Discrete Probability Distribution,The quantity P(X=xi) is denoted by f(xi) for all i and is called the probability distribution function or probability distribution of the random variable X. For example,51,2.5 Discrete Probability Distribution,Orders for General Avionics 6

36、36 Aircraft Probability xi pi 42 0.05 43 0.10 44 0.15 45 0.20 46 0.25 47 0.15 48 0.10,p1+p2+pn = 1,52,2.5 Discrete Probability Distribution (3),Graph of the probability distribution of the number of orders for General Avionics 636 aircraft next year,A histogram is a display of probabilities as a bar

37、 chart,53,Discrete Random Variables,A probability distribution for a discrete random variable X consists of (i) possible values x1, x2, . . . , xn, (ii) corresponding probabilities p1, p2, . . . , pn, with the interpretation that P(X = x1) = p1, P(X = x2) = p2, . . . , P(X = xn) = pn,Note: (1) Rando

38、m variable is denoted by (X) (2) Possible outcomes denoted by (x1, x2, . . . ) (3) Probabilities must sum to 1.0 : p1 + p2 + . . . + pn = 1.0 pi 0,54,Where do probability distributions come from?,2Theoretically,I:Empirically (from data),55,I:Empirically (from data),Example: The Kentucky Fried Chicke

39、n “restaurant” in Porter Square sells chicken in “buckets” of 2, 3, 4, 8, 12, 16 or 20 pieces. Over the last week, orders for fried chicken had the following data:,Let X = number of pieces of chicken in an order. Develop a probability distribution for X.,56,2Theoretically,The Binomial Distribution 二

40、项分布,The experiment consists of n “trials”,Each trial results in : “success” or “failure”,Trials are independent of each other,Each trial has same probability: success p, failure 1-p X= # successes in n trials.,Random variable X has binomial distribution with parameters n and p “Binomial(n,p)”,57,Exa

41、mple 1: 70% are US, and 30% are International,X = number of international students on the DMD team,n = 3 trials,an international student “success” a US student “failure”,P(success) = p = 0.30; P(failure) = 1 - p = 0.70,The Binomial Distribution,58,The Binomial Distribution,P(X=0) =,P(X=3) =,+ 0.7*0.

42、3*0.7,P(X=1) =,= 3*(0.7) 2 *(0.3) = 0.441,0.7*0.7*0.7 =(0.7) 3 = 0.343,0.3*0.3*0.3 =(0.3) 3 = 0.027,0.3*0.7*0.7,+ 0.7*0.7*0.3,59,Let X be Binomial(n, p),Check in Example 1: n=3 , p=0.3,= * 0.7 2 * 0.3,P(X=1)=?,3,3!/(2!1!),P(X=x)=,px (1-p) n-x,60,例2.8 生产过程中的质量问题 假设一个生产过程生产出的产品是优质产品的概率是0.92,是劣质产品的概率是0

43、.08。假设此生产过程每次生产140单位的产品。用X表示每次生产的优质产品数。则X服从二项分布,样本容量n=140,发生概率p=0.92。注意X可以取0,1,2,3,140中的任何值,取到不同的值的概率也不同。 例2.9 掷硬币 另一个服从二项分布的随机变量是掷n次硬币出现正面的次数。用H表示掷10次硬币出现正面的次数。则H服从二项分布,样本容量n=10,发生概率p=0.5。,61,Example 2.10- Computers in the Office There are 8 computers in a small retail office. The probability of an

44、y one computer being used during the lunch hour is 0.70. Let X denote the number of computers in use during the lunch hour. Then X obeys a binomial distribution with sample size n=8 and probability of success p=0.70.,2.6 The Binomial Distribution,62,Example 2.11- Quality Control in a Production Proc

45、ess The Production process produces 5 units per shift. A unit produced by the production process is of good quality with probability 0.83 and is of poor quality with probability 0.17. Let X denote the number of units that are of poor quality in one shift, then X obeys binomial distribution with n=5

46、and p=0.17. 这批部件中,正好有2个部件是质量差的概率;,2.6 The Binomial Distribution,计算在一批中至少有一个部件是质量差的概率。,63,例2.12 Netway Computers公司服务中心的人员安排问题,Frank是Netway Computers公司Seattle地区的区域经理。过去几年中Netway Computers公司在Seattle地区卖出了50台计算机,这样就需要做相应的售后服务。 Frank需要决定在服务中心应当安排多少名工程师,才能以一个很大的概率来保证客户的要求能够得到满足。,64,历史情况与要求的服务条件,从历史情况来看,Fra

47、nk估计在一天内一个客户需要服务的可能性为0.02。很显然,工程师越多,客户得不到服务的可能性就会越低。Frank想安排一定数目的工程师使得客户得不到服务的可能性至多为5%。,65,概率模型,设X为给定的某一天请求服务的客户数量,为随机变量。 在任意一天,一个客户需要服务的可能性为p=0.02。 X是一个服从二项分布的随机变量,其中n=50,p=0.02。,66,若安排一名工程师,若在某一天,如果至少有2个客户请求服务,那么至少有一个客户没有得到服务,这个事件的概率为 P(X 2) 计算该事件的概率, P(X 2)=1-P(X=0)- P(X=1) 利用二项分布公式进行计算 P(X 2)=1-

48、P(X=0)- P(X=1)=1-0.364-0.371=0.265 因此,在安排一名工程师的情况下,至少有一个客户得不到服务的概率为0.265。,67,Answer,同理,P(X3)=0.080,P(X4)=0.020 安排2位工程师时,至少有一个客户得不到服务的可能性为0.08,即8%,这还不能满足小于5%的要求; 安排3位工程师的时候,至少有一个客户得不到服务的可能性为0.020,即2%。因此安排3个工程师就能够达到Frank对服务质量保证的目标,68,Uncertain environment How to compare random variables X and Y?,69,EX

49、AMPLE,X and Y denote the sales next year in the eastern division and the western division of a company, respectively. X and Y obey the following distributions:,Eastern Division,Western Division,70,2.7 Summary Measures of Probability Distributions,Main Concepts: Mean or expected value of a random var

50、iable Variance of a random variable Standard deviation of a random variable Coefficient of variation of a random variable,71,Mean or Expected Value: Represents “average” outcome; a measure of “central tendency” 集中趋势,Variance: Squared deviation around the mean; a measure of “spread” 离散程度,Standard Dev

51、iation : Measured in the same units as the random variable X. 例如学生的平均成绩以及方差, 员工工资。(所有数据信息or极端数据) 集中趋势:中位数(Median 奇数、偶数) 众数(Mode,没有或者多个),Three important measures of a random variable,0,8,12,20 8,9,11,12,均值 10,72,Example: Let X be the number that comes up on a roll of a die.,Outcome P(X=x) 1 1/6 2 1/6

52、 3 1/6 4 1/6 5 1/6 6 1/6,1/6*1 +,1/6 *2 +,1/6 *3 +,1/6*4,= 3.5,1/6*(1-3.5)2 +,1/6 *(2-3.5) 2 +,1/6 *(3-3.5) 2 +,Sqrt2.917= 1.708,+1/6*5,+ 1/6*6,1/6 *(4-3.5) 2 +,1/6 *(6-3.5) 2,1/6 *(5-3.5) 2 +,= 2.917,73,Example 1: 30% of first-year Sloan students are “international students.” Professors Freund, Per

53、akis, and Wang each randomly select one of their students to participate in the DMD curriculum assessment team. Let X = number of international students (Int) on the DMD team.,0.343*0 +,0.441 *1 +,0.189 *2 +,0.027 *3,= 0.9,0.343*(0-0.9)2 +,0.441 *(1-0.9) 2 +,0.189 *(2-0.9) 2 +,0.027 *(3-0.9) 2,= 0.6

54、3,Sqrt0.63= 0.79,74,Example 2.15-Sales Performance in Two Regions,2.7 Summary Measures of Probability Distributions,Eastern division Sales ($million) Probability 3.0 0.05 4.0 0.20 5.0 0.35 6.0 0.30 7.0 0.10,Western division Sales ($million) Probability 3.0 0.15 4.0 0.20 5.0 0.25 6.0 0.15 7.0 0.15 8.

55、0 0.10,Let X and Y be the sales of the product in the eastern and western sales division for the next year respectively, then,E(X)=0.05(3.0)+0.20(4.0)+0.35(5.0)+0.30(6.0)+0.10(7.0)=$5.2 million E(Y)=0.15(3.0)+0.20(4.0)+0.25(5.0)+0.15(6.0)+0.15(7.0)+0.10(8.0)=$5.25 million,75,Example 2.15-Sales Perfo

56、rmance in Two Regions,2.7 Summary Measures of Probability Distributions (5),Probability distribution of eastern sales X,Probability distribution of western sales Y,E(X)=$5.2 million,E(Y)=$5.25 million,2X =0.05(3.0-5.2)2+0.20(4.0-5.2)2+0.35(5.0-5.2)2+0.30(6.0-5.2)2+0.10(7.0-5.2)2=1.06 2Y =0.15(3.0-5.

57、25)2+0.20(4.0-5.25)2+0.25(5.0-5.25)2+0.15(6.0-5.25)2+0.15(7.0-5.25)2 +0.10(8.0-5.25)2=2.3875,76,二项分布的均值和标准差,如果X是一个服从二项分布的随机变量,样本容量为n,成功概率为p,则其均值、方差和标准差分别为:,77,Cases that coefficient of variation is more suitable for measure of variability: When two populations of very different nature are to be comp

58、ared, e.g., the variability of height and the variability of weight; 量纲不同 均值不同 When two populations of same nature with great difference in mean values are to be compared, e.g., packages of weight 50 grams and packages of weight 500 grams. 数量级差别大,2.7 Summary Measures of Probability Distributions,Coefficient of variation(变异系数) of a rand

温馨提示

  • 1. 本站所有资源如无特殊说明,都需要本地电脑安装OFFICE2007和PDF阅读器。图纸软件为CAD,CAXA,PROE,UG,SolidWorks等.压缩文件请下载最新的WinRAR软件解压。
  • 2. 本站的文档不包含任何第三方提供的附件图纸等,如果需要附件,请联系上传者。文件的所有权益归上传用户所有。
  • 3. 本站RAR压缩包中若带图纸,网页内容里面会有图纸预览,若没有图纸预览就没有图纸。
  • 4. 未经权益所有人同意不得将文件中的内容挪作商业或盈利用途。
  • 5. 人人文库网仅提供信息存储空间,仅对用户上传内容的表现方式做保护处理,对用户上传分享的文档内容本身不做任何修改或编辑,并不能对任何下载内容负责。
  • 6. 下载文件中如有侵权或不适当内容,请与我们联系,我们立即纠正。
  • 7. 本站不保证下载资源的准确性、安全性和完整性, 同时也不承担用户因使用这些下载资源对自己和他人造成任何形式的伤害或损失。

评论

0/150

提交评论