251y9872 11/27/98 ECO251 QBA1 THIRD EXAM NOVEMBER 24, 1998 Name __________________ Section Enrolled: MWF TR 10 11 12:30 Part I. Do all the Following (10+ Points) Penalty for not doing question 1. Show your work! You are following a stock and wish to compare its return to the Dow-Jones index. The return for 9 periods on the stock ( y ) is compared below to the return on “buying the Dow”( x ). Period x (Dow) 1 2 3 4 5 6 7 8 9 Total y (Stock) 12 6 2 4 4 5 6 -6 3 36 12 15 -4 1 2 -1 -8 -2 3 18 y2 144 225 16 1 4 1 64 4 9 468 xy 144 90 -8 4 8 -5 -48 12 9 206 Note that x 4, s x 4.7170 Compute the following: 1. The sample variance for the stock (4). 2. The sample covariance between the stock and the Dow (2) 3. The sample correlation between the stock and the Dow (2) 4. Interpret the correlation (1) Answers to questions 5) and 6) must be based on the results in questions 1-4. Do not recompute the answers after changing y ! 5. If all the numbers in the y column were higher by 3 (i.e. 15, 18, -1, 4 etc.), what would the variance, covariance and correlation computed above be? (1.5) 6. If all the numbers in the y column were twice as high (i.e. 24, 30, -8, 2 etc.), what would the variance, covariance and correlation computed above be? (1.5) Solution: 1) y 18 2 , y 2) n 9 xy 196 , s xy sx s y y 2 ny 2 n 1 468 922 54 8 ( s y 54 7.34847 ) xy nx y 206 942 16.75 . n 1 9 1 16 .75 0.4832 4.7170 54 4) The positive sign of rxy , the sample correlation, indicates that x and y tend to move together. If we 3) rxy s xy s 2y square rxy , we get approximately .23, which on a zero to one scale indicates a relatively weak relationship. 251y9872 11/27/98 5) We are leaving x alone, but replacing y by y 3 . From the syllabus supplement and the outline if v cy d y 3 so that c 1 and d 3 a) v2 Varv c2Var y c2 y2 12 y2 54 b) If w ax b and v cy d , wv Cov(ax b, cy d ) acCov( x, y) and a 1 and b 0 wv acCov( x, y) 1116.75 16.75 . c) Corrw, v rwv signacrxy sign11.4471 sign 1.4832 .4832 6) We are leaving x alone, but replacing y by 2 y . From the syllabus supplement and the outline if v cy d 2 y so that c 2 and d 0 a) v2 Varv c2Var y c2 y2 22 y2 454 216 b) If w ax b and v cy d , wv Cov(ax b, cy d ) acCov( x, y) and a 1 and b 0 wv acCov( x, y) 1216.75 33.50 . c) Corrw, v rwv signacrxy sign12.4471 sign 2.4832 .4832 2 251y9872 11/27/98 Part II. Do the following problems ( do at least 40 points ). Show your work! Note: You only need 40 points out of the 72 below to get an A+ -. Do the parts that look easiest! 1. The following table represents the joint probability of x and y . Event x 2 y 3 4 Px 0 P y yP y y 2 P y 1 2 .25 .10 .05 .25 .15 .10 .05 .05 ? .40 .80 .35 1.05 .25 1.00 1.60 3.15 4.00 .10 1.00 2.85 8.75 .45 .45 So E y y 2.85 and E y 2 8.75 0+ .45 + .20 = 0.65 Ex x xPx x P x 0+ .45 + .40 = 0.85 E x 2 2 a. b. Fill in the missing number. (1) Are x and y independent? Why?(2) c. Compute xy , the covariance of x and y , and interpret it. (3) d. Compute xy , the correlation of x and y , and interpret it. (3) e. f. Find the probability that x y is less than 6. (1) (i) Find the distribution of x y . (2) (ii) Using only the results of a)-d), find the mean and variance of x y .(3) Solution: a) b) c) Since the table must total 1.00, the missing number is 0. x and y are independent if Px, y Px P y . In the lower right corner 0 10 .25 , so x and y are not independent . E x, y xyPx, y xy Covx, y Exy x y 1.85 0.65 2.85 0.0025 Negative, so x and y move in opposite directions. = 0(2)(.25) + 1(2)(.10) + 2(2)(.05) + 0(3)(.05) + 1(3)(.25) + 2(3)(.05) + 0(4)(.15) + 1(4)(.10) + 2(4)(0) = 1.85 d) xy xy .0025 Ey .0025 2 0.4275 0.6275 0.0000232 0.0048 0.4275 0.6275 We measure the strength of a correlation by squaring it. If we square -.0048, we get .000023. On a zero to one scale, this is tiny, so correlation is very weak. x y Var x x2 E x 2 x2 0.85 0.65 2 0.4275 Var y y2 e) 2 2 y x 0.4275 0.6538 8.75 2.85 0.6275 y 0.6275 0.79215 2 See below. Since the sums of x and y are all below 6, Px y 6 1 3 251y9872 11/27/98 sums probabilit ies x x 0 1 2 0 1 2 f-i) 2 2 3 4 2 .25 .10 .05 y 3 3 4 5 y 3 .05 .25 .05 4 4 5 6 4 .15 .10 0 The tables above show the sums of x and y and their probabilities . If we add the probabilities we find the distribution below. P x y x y 3 P2 0 P0 3 1 2 4 P0 4 1 3 2 2 2 5 6 P1 4 2 3 P2 4 x y Px y x y 2 P x y .25 = .25 .50 1.00 .05 +.10 = .15 .45 1.35 .15 +.25+.05 = .45 1.80 7.20 .10 +.05 = .15 .75 3.75 0= .00 .00 0.00 1.00 3.50 13.30 Only the x y and Px y columns are needed. From this, Ex y 3.50 and Varx y E x y 2 x y Ex y and Var x y . 2 13 .30 3.50 2 3.05 . But this is not how I asked you to compute f-ii) x y Ex y x y Ex E y 0.65 2.85 3.50 Varx y Varx Var y 2Covx, y 0.4275 0.6275 2(0.0025 ) 1.05 4 251y9872 11/27/98 2. I sell both coal and oil. Although the prices of oil and coal fluctuate, I maintain a constant markup of $16.00 per ton for coal and $0.06 per gallon on oil. My fixed costs for coal storage are $600 a month. My fixed costs for oil storage are $300/month. My mean sales of coal are 3000 tons, with a standard deviation of 50 tons. My mean sales of oil are 4500 gallons with a standard deviation of 300 gallons. The correlation between sales of coal and oil is –0.3. a. What are the mean and standard deviation of profits on coal? (3) b. What are the mean and standard deviation of profits on oil? (1) c. What is the covariance between sales of coal and oil? (1.5) d. What is the covariance between profits on coal and oil? (1.5) e. What is the mean and standard deviation of total profits (oil profits plus coal profits)? (4) Solution: If coal is x and oil is y , x 3000, x 50, y 4500, y 300 and xy 0.3 . a) Let profits on coal be w . From the syllabus supplement and the outline if w ax b 16 x 600 so that a 16 and b 600 , then: E w E ax b aEx b or E16 x 600 16 Ex 600 163000 600 47400 w2 Varw a 2Varx a 2 x2 16 2 x2 16 2 50 2 640000 , w 640000 800 b) Let profits on oil be v . From the syllabus supplement and the outline if v cy d .06 y 300 so that c .06 and d 300 , then: E v E cy d cE y d or E.06 y 300 .06 E y 300 .064500 300 30 . v2 Varv c 2Var y c 2 y2 .062 y2 .062 3002 324, v 324 18 c) Cov( x, y) xy xy x y 0.350300 4500 d) Cov( x, y) wv Cov(ax b, cy d ) acCov( x, y) 16 .06 4500 4320 e) From the previous page substituting w for x and v for y , x y Ew v w v Ew Ev 47400 30 47370 Varw v Varw Varv 2Covw, v 640000 324 2(4500 ) 631324 wv 631324 794 .559 5 251y9872 11/27/98 3. Assume that my father gives me a present of $1000 every time my wife has a boy baby and $2000 every time she has a girl baby. (Assume that the chance of a girl baby is 50%) a. What is the chance that out of 10 births exactly 3 will be girls? (2) b. What is the chance that out of 10 births more than half will be girls? (2) c. What is the mean and variance for the amount of money that I receive on any one birth? (2) d. What is the mean and variance for the number of girls born in 10 births? (1.5) e. What is the chance that the first girl will be born on the 10 th birth ? (1.5) If a variable has the binomial distribution with p .03 and n=10 (i) Find the probability that you have at least one success (2) (ii) Is it appropriate to use the Poisson distribution in this problem? Why? If it is appropriate, use the Poisson distribution to find the answer to (i). (2) f. Solution: Binomial n 10, p .5 . (Because we are asking for the probability of x successes in n tries, and the probability of success on any one try is given and constant.) a) b) c) P3 Px 3 Px 2 .17188 .05469 .11719 , also C319 .53 .57 . Px 5 Px 6 1 Px 5 1 .62305 .37695 y P y yP y y 2 P y So y 1500 and y2 250000 15002 250000 .5 500 .5 1000 1.0 1500 500000 2000000 2500000 y 500 1000 2000 d) np 10.5 5.0 2 npq 10.5.5 5.0.5 2.50 e) Geometric p .5 (Because we are asking for the probability of the first success on try x .) P10 q 9 p .59 .5 .0009766 f) (i) Binomial n 10, (ii) Since 0 10 10 p .03 . Px 1 1 P0 1 C 010 p q 1 .97 1 .73742 26258 n 10 333 .33 p .03 is less than 500, the Poisson distribution is not appropriate. 6 251y9872 11/27/98 4 . . Make sure that you state clearly what distribution you are using in each section of this problem. If I find that my paint jobs have an average of 0.10 blisters per square foot and a refrigerator has 14 square feet of painted surface, a. What is the chance that there will be exactly one blister on a refrigerator? (1) b. What is the chance that there will be at least one blister on a refrigerator? (2) c. What is the standard deviation of the number of blisters on a refrigerator? (1) d. Using your results from a), what is the chance that, in a delivery of 10 refrigerators, all will have at least one blister? (2) e. In d) what is the chance that more than 8 will have at least on blister? (2) f. If a store sells an average of 72 refrigerators in a 16 hour day and it takes 2 hours to get a delivery, what is the fewest number of refrigerators it should have in stock before it reorders if it wishes to limit the chance of running out to not more than 1% ? Explain. (3) Solution: Poisson with parameter (mean) of 0.1(14) = 1.4. (Because we are asking for the probability of x successes when the average number successes in a unit of space or time is given and constant.) a) P1 Px 1 .345236 b) Poisson Px 0 Px 1 1 P0 1 .246597 .753403 c) d) Since, for the Poisson distribution, the mean and variance are identical, m 1.4 1.1832 Binomial n 10, p .753403. (Because we are asking for the probability of x successes in n tries, and the probability of success on any one try is given and constant.) Px C xn p x q n x so e) 10 .753403 10 1 .753403 0 .753403 10 .058921 P10 C10 Binomial n 10, p .753403 Since P9 C910 .753403 9 1 .753403 1 10.753403 9 .241597 10.018894597 .1889460 , Px 8 P9 P10 .1889460 .058921 .247867 f) 72 9. 8 From the Poisson (9) table, the first value of k with Px k .99 is 17 with Px 17 .99468 . So the reorder point is 15. Since there are 8 two-hour periods in a 16 hour day, this is a Poisson problem with m 7 251y9872 11/27/98 5. Assume that in any batch of soft drink cans 5% are defective. Make sure that you state clearly what distribution you are using in each section of this problem. a. If I take 4 cans from a package of 20 cans what is the chance that exactly 2 are defective? (2) b. If I take 4 cans from a package of 20 cans, what is the chance that at least one is defective? (2) c. What are the mean and variance of the number of defective cans in a sample of 4 taken from a package of 20? (2) d. If I take 4 cans from a package of 2000 cans, what is the chance that exactly 2 are defective? (2) e. If I take 4 cans from a package of 2000 cans, what is the chance that at least one is defective? (2) f. What are the mean and variance of the number of defective cans in a sample of 4 taken from a package of 2000? (2) g. If I am taking cans from a package of 2000, what is the chance that I will find my first defective can in the first 100 cans (2) ? Solution: Hypergeometric with N 20, M pN .05 20 1 and n 4 (Because we are asking for the probability of x successes when the population is of limited size and the number of successes in the population is also limited.) a) Since M 1, P2 0 b) Since Px C xM C nNxM C nN , Px 0 1 P0 1 C 01 C 419 C 420 19! 19 18 17 16 15!4! 1 4 3 2 1 1 19 18 17 16 1 16 1 .80 .20 1 20! 20 19 18 17 20 20 19 18 17 16!4! 4 3 2 1 1 N n 20 4 16 npq 4.05 .95 0.1900 .1600 N 1 20 1 19 d) Though hypergeometric is not wrong here, because of the large population we are better off using the binomial. So the distribution is either Binomial with n 4, p .05 or Hypergeometric with c) np 4.05 0.20 2 N 2000 , M pN .05 2000 100 and n 4 . P2 Px 2 Px 1 .99952 .98598 .01354 or 100 99 1900 1899 100 99 1900 1899 4 3 2 1 2 1 P2 .01344 2000 2000 1999 1998 1997 2000 1999 1998 1997 2 1 C4 4 3 2 1 Px 0 1 P0 1 .81451 .18549 or C 2100C 21900 e) Px 0 1 P0 1 C 0100C 41900 C 42000 2 1 1900 1899 1898 1897 1 .8143775 .18562 2000 1999 1998 1997 f) np 4.05 0.20 npq 4.05.95 0.1900 g) Geometric p .05 (Because we are asking for the probability of the first success on try x .) Remember F x 1 q x where q 1 p . Px 100 F 100 1 q100 1 .95 100 1 .0059 .9941 8 251y9872 11/27/98 6. A jorcillator has two components, a phillinx and a flubberall. As long as both are working the jorcillator will not be junked. If either fails the jorcillator must be junked. a. If a phillinx has a lifespan approximated by a uniform distribution between 4 and 14 years, find: (i) The mean and standard deviation of its life. (2) (ii) The probability that it will last between 0 and 5 years (2) (iii) The probability that it will last between 5 and 10 years (1) (iv) The probability that it will last more than 10 years (1) b. If a flubberall has a lifespan approximated by a uniform distribution between 5 and 20 years, do (i) a(ii)-a(iv) above for the flubberall (2) and then (assuming that the lifespan of the two components is independent) find: (ii) The probability that the jorcillator will last between 0 and 5 years. (2) (iii) The probability that the jorcillator will last between 5 and 10 years. (3) (iv) The probability that the jorcillator will last more than 10 years. (2) Solution: a) For the Phillinx: c 4, d 14 c d 4 14 9 2 2 d c2 14 42 8.3333 2 12 12 (i) 8.3333 2.8868 For the logic of the rest of this problem, see problem H4. 1 1 P A P0 x 5 5 4 10 10 =.100 1 5 PB P5 x 10 10 5 10 10 =.500 1 4 PC Px 10 14 10 10 10 =.400 (ii) (iii) (iv) Phillinx (Draw a diagram for the uniform distribution) Flubberall (Draw a diagram for the uniform distribution) b) For the Flubberall: c 5, d 20 (i) PD P0 y 5 0 PE P5 y 10 10 5 1 1 =.3333 15 3 1 2 PF P y 10 20 10 =.6667 15 3 b-ii) A joint probability table follows on the left with the unions of joint events that make up the events that were specified summed to the right. Note that the probabilities sum to one. D A 0 B 0 C 0 0 E 1 30 5 30 4 30 1 3 F 2 30 10 30 8 30 2 3 1 10 5 10 4 10 1 Event Fails in 0-5 Fails in 5-10 Fails in 10+ Sum Component Joint Events P A D A E A F B D C D PB E B F C E PC F Probability 1 2 3 0 00 .10000 30 30 30 5 10 4 30 30 30 8 30 19 .63333 30 8 .26667 30 1.0000 9