PoissonAndExponential
3 Pages

PoissonAndExponential

Course Number: MATH 218, Spring 2000

College/University: USC

Word Count: 903

Rating:

Document Preview

Poisson and Exponential, Connected The Poisson distribution counts the number of discrete events in a fixed time period; it is closely connected to the exponential distribution, which (among other applications) measures the time between arrivals of the events. The Poisson distribution is a discrete distribution; the random variable can only take nonnegative integer values. The exponential distribution can take any...

Unformatted Document Excerpt
Coursehero >> California >> USC >> MATH 218

Course Hero has millions of student submitted documents similar to the one
below including study guides, practice problems, reference materials, practice exams, textbook help and tutor support.

Course Hero has millions of student submitted documents similar to the one below including study guides, practice problems, reference materials, practice exams, textbook help and tutor support.

and Poisson Exponential, Connected The Poisson distribution counts the number of discrete events in a fixed time period; it is closely connected to the exponential distribution, which (among other applications) measures the time between arrivals of the events. The Poisson distribution is a discrete distribution; the random variable can only take nonnegative integer values. The exponential distribution can take any (nonnegative) real value. The Poisson process is completely described by a parameter , the average number of arrivals per unit time. If X denotes the number of events which happen in unit time, then k - e . k! This is connected to the random variable Y , the time between successive events, by the fact that Y is exponentially distributed with density function e-x , and therefore P (X = k) = t (1) (2) P (Y t) = 0 e-x dx = 1 - e-t , P (Y > t) = 1 - P (Y t) = e-t . Sometimes we may wish to speak about the number of arrivals in a time interval T which is different from the unit time interval. This is done by scaling: multiply (average number of arrivals per unit time) by T (number of units of time) to get T , the average number of arrivals in time T . (So if an average of six babies arrive per day in a hospital, then an average of 0.25 babies arrive per hour.) The probability that exactly k events will occur in time interval T is therefore (3) (T )k -T e . k! This makes the connection between the Poisson and exponential distributions transparent: for the event Y > T means that the first arrival takes more than T units of time; which means that there are 0 arrivals in the first T units of time. Taking k = 0 in (3), this has probability e-T , i.e. P (Y > T ) = e-T . This was exactly what we claimed for the exponential distribution in (2). For an example, we will model the arrival of babies at a lying-in hospital. These are Poisson events, because they satisfy the three characteristics of a Poisson model: Arrivals of babies are independent events. (The mothers don't conspire together.) The probability of babies being born almost simultaneously is negligible. The probability of a baby being born in a time interval only depends on the length of the time interval, not on when it is during the day. For many years the hospital has been averaging 6 births per day. That means that the average time between births will be 1/6 of a day. Therefore if X denotes the number of babies born in a day, then 6k P (X = k) = e-6 , k! 1 2 while if Y denotes the time between births, then P (Y > t) = e-6t . The probability density function for Y is 6e-6x . Recall that the exponential which distribution has PDF e-x has mean 1/, so this fits: Y should have mean 1/6 (days/birth). Also note how easy it is to connect these parameters through "dimensional analysis": X, the number of births, is measured by babies/day; Y , the time between births, is measured in days/baby. So 6 babies/day is equivalent to 1/6 days/baby, or 4 hours/baby. For example, in the above scenario, given that a baby has just been born, the probability that the next baby will take one or more hours to arrive is found by converting hours to days, 1 hour = and computng P (Y > Surely time enough for a coffee break! Consistency of the Poisson model. How far can we trust the Poisson model? Actually, pretty far. We'll give an example to illustrate how consistent it is. It illustrates a more general fact (the sum of two independent Poisson RV's with mean and is a Poisson RV with mean + ). This is quite reasonable: think of two hospitals, one giving birth to an average of babies per day, the other giving birth to babies per day. If you combine the hospitals (say, a health conglomerate buys them both) then the conditions for a Poisson RV are satisfied for births at the combined hospitals, which would average + per day.) Consider a hospital which has an average of 6 babies born each day. There are two shifts for the nursing staff, so the administration prefers to use 12-hour periods, with an average of 3 babies born each 12-hour period. Does using different scales like this make a difference in the probabilities? No, if we keep the time intervals strai...

Find millions of documents on Course Hero - Study Guides, Lecture Notes, Reference Materials, Practice Exams and more. Course Hero has millions of course specific materials providing students with the best way to expand their education.

Below is a small sample set of documents:

Maryland - ENEE - 498
%!PS-Adobe-2.0 %Creator: dvips(k) 5.86 Copyright 1999 Radical Eye Software %Title: lasergame.dvi %Pages: 7 %PageOrder: Ascend %BoundingBox: 0 0 612 792 %DocumentFonts: CMBX12 CMR12 CMTI12 CMR10 CMMI12 CMMI10 CMSY10 CMR6 %+ CMR8 %EndComments %DVIPSWeb
USC - MATH - 126
MIDTERM 1 SOLUTIONS MATH 126, SEPTEMBER 24, 2003Problem 1. (20%) Evaluate, showing all steps: 1 1 - ln(1 + x) sin x (b) lim e-x x3 x ex - 1 (c) lim x0 sin x (a) limx0the time you finish this course you should be able to argue as follows: we have
USC - MATH - 425
SOLUTIONS TO FINAL EXAM, MATH 425A GIVEN MONDAY, DECEMBER 15, 2003Putting (1) and (2) together, we have d(x, y) = max(d(x, z), d(z, y), and therefore d(x, y) is either equal to d(x, z) or equal to d(z, y). That is, the triangle has two equal sides.
USC - MATH - 425
MIDTERM 1 SOLUTIONS MATH 425A, OCTOBER 15, 2003Proof. Call the shued sequence {zn }; that is, dene zn = yk , xk , n = 2k; n = 2k 1.Problem 1. (Quickies) For each statement below, determine whether it is true or false and explain why. Explanation
Maryland - ENEE - 611
ENEE 611: Homework on LayoutsRemember that the pictures from the lecture, including color versions of the pictures in this homework, can be found at: http:/www.ece.umd.edu/ dilli/courses/enee611/layoutlecture/index.html1. Draw the cross-sections al
USC - MATH - 218
0.175 0.15 0.125SECOND MIDTERM, MATH 218 NOVEMBER 9, 20050.1 0.075 0.05 0.025 0 8.5 9 9.5 10 10.5 11 11.5Problem 1. (10%) A baker sells cakes. (Mmmm, mmmm, good!) The bakers monthly prot from the sale of cakes is normally distributed with a mea
USC - MATH - 218
Raw Scores for Math 218, Section 50342, Fall 2005Code bax buf bup byv cap ceh cov cyn ded fem m fur fuv gag gyj haj hon hox hud huv jaq jex kod kov kyk law leq lif lolMid1 84 68 78 96 52 83 85 92 92 88 87 93 64 91 81 99 83 84 81 89 65 76 94 90 78
Washington - STAT - 582
o 0.20 o Error o o o o 2 4 x o oo o o o o o o o o o o o o o o o o o o o o oo o ooo o o oo o o oo oo oo 20 30 40 x 50 60 6 8 10 -0.005 0 0.005 o oProbability0.10 o o 0.0 o 024 x6810Probability0.060.02
Washington - STAT - 582
dnorm(x) * (x^4 6 * x^2 + 3) 0.5 0.0 3 2 1 dnorm(x) * (x^6 15 * x^4 + 45 * x^2 5) 2 3 2 1 0 1 2 3 x x 0 2 4 6 3 0 x 1 2 3 3 2 1 0 1 2 3 2 1 0.3 0 x 0.1 1 0.1 2 dnorm(x) * (x^3 2 * x) 0.3 3 0.5 1.0 0.4 0.2 0.0 0.2dnorm(x) * (x^2 1)
USC - MATH - 218
Math 218, Fall 2005, 10:0010:50MWF Class Number 50342DThis document is distributed to the class in printed form, but the most current is always available on the World Wide Web, at URL <http:/imperator.usc.edu/bruck/classes/fall2005/math218.09/>Pro
Washington - STAT - 582
STAT 582Homework 1.Due Wednesday, January 14. 1. Let X1,K, X n be iid U (q - 1 2 ,q + 1 2). Let T = (U,V ) = (minX i ,maxX i ) and S = V - U. (a) Derive the conditional distribution of T given that S = s. (b) Show that a 95% conditional confidence
Washington - STAT - 582
STAT 582Winter 2004HOMEWORK SET 2Due January 21 1. Consider a situation with sufficient statistic(T,A), where T is the mle of q and A is ancillary for q. (a) Show that the Fisher information about q from observing T only is no larger than the Fi
Washington - STAT - 582
STAT 582Winter 04Homework 3Due January 28 1. (a) Let f X ,Y (x, y;q ) = exp(-qx + y /q ), x, y > 0. Based on a sample of size n find a sufficient statistic for q, the mle for q, and the Fisher information about q in the sample. (b) Find the dens
Washington - STAT - 582
STAT 582Winter 2004Homework 4.Due February 4 1. Show that the two first correction terms in the Edgeworth expansion of the mean of N(0,1)-random variables vanish. 2. (a) Do an Edgeworth expansion up to terms of order n -1 for the distribution of
Washington - STAT - 582
Severini problem 2.6 Consider a random variable with mean 0 and variance 1. Suppose the density of the random variable is approximated by an Edgeworth expansion including only the 1/ n term. Using this approximation, find approximations for the media
Washington - STAT - 582
STAT 582Winter 04Homework 5Due February 11 Problems 2.7 2.12 (pp 70-71) in Severini.
Washington - STAT - 582
STAT 582Winter 2004Homework 6.Due February 27 (note due on Friday) 1.Let T(F) = F 1 ( 1 2) = inf{x : F(x) 1 2}. Assume that f (F 1 ( 1 2) > 0. (a) Show that F has a unique median. (b) Show that if f > 0 we have, writing F = F + (G F), (1 )F(F
Washington - STAT - 582
STAT 582W04Homework 7.Due Monday March 82 1. Let T(F) = F = ( E F X ) . Using the asymptotic theory for statistical functionals, find the limiting distribution of T(Fn ) when F = 0. x t 2. Let X1,., X n be iid nonnegativee random variables with
Washington - STAT - 582
STAT 582Winter 04Solution, homework 31.(a) The likelihood is L( ) = exp(-( x i +1 y i ) , so we have y ^ ( ) = - x + 1 y and = i . By the factorization theorem X , Y is l i i i i 2 xi ( ) = - 2 y so I( ) = 2 nE(Y ) = 2n . sufficien
Washington - STAT - 582
Stat 582Winter 04Solution, homework 41. By symmetry 3 = 0. The fourth central moment is 3, so the kurtosis coefficient 4 3 is also 0. Hence there is no correction, and the two-term Edgeworth expansion is in fact exact. 2. The Poisson mgf is M(t
Washington - STAT - 582
dpois(x, 15) - spa -1e-04 0 10 20 30 40 Index 40 0 10 20 30 30 -0.002 20 Index Index 0.001 10 0.003 0 0e+00 1e-04 -1e-04 0e+00 1e-04dpois(x, 15) - edgedpois(x, 15) - dnorm(2 * (sqrt(x) - sqrt(15)/sqrtEdgeworth error Varstab errorn = 5, lambda
Washington - STAT - 582
STAT 582Winter 04Solution, homework 51. EX n = 1 and VarX n = 1 2 , so by CLT Fn (x) ( n (x -1). n x r +n-1 (n ) n (n + r) r exp(-xn )dx = (n )-r 2. (a) X n ~ (n,n ) so EX n = . Using (n) (n) 1 1 1 + + O(x -3 ) and taking logs we get Stirli
Washington - STAT - 582
STAT 582W04Solution, homework 6.1. (a) Since F is montonone increasing at F-1(1/2) the median is unique. (b) Assume that G is in the same class as F, i.e. has positive density. 1 G(F 1 ( 1 ) 1 G(F 1 ( 1 ) 2 2 F1 ( 1 2) = inf{x : F (x) 1 2}
Washington - STAT - 582
STAT 582W 04Solution, homework 71. T(F + (G - F) = (F + (G - F ) 2 , so d1T(F;G - F) = 2(G - F )F = 0. 2 d2T(F;G - F) = 2(G - F ) 2 = 2G = 2 xydG(x)dG(y). Hence h(F;x, y) = xy - ydF(x) - xdF(x) + xydF(x)dF(y) = xy. Checking the regular
Washington - STAT - 582
STAT 582 Name:_W03Final examPlease write complete solutions, and remember to put your name on each additional piece of paper you use.c(q )exp(- qx 2 ) x 0 1. The density for X given q is f (x | q ) = . 0 x<0 (a) What is the natural parameter
Washington - STAT - 582
STAT 582FINAL EXAMMarch 11, 19961. Let X 1 ,X 2 ,X 3 given W=w be iid Bern(w). You observe X 1 , X 2 , and Y=1(X 1 +X 2 =0) X 3 . (a) Prove or disprove: S=X 1 +X 2 +Y is a sufficient statistic. (b) Find the Bayes estimate for W with squared err
USC - MATH - 218
0.175 0.15 0.125 0.1 0.075 0.05 0.025 0 5 6 7 8SECOND MIDTERM, MATH 218 APRIL 7, 2006Problem 1. (15%) An auto dealer's monthly profit from the sale of cars is normally distributed with a mean of $40000 and a standard deviation of $3000. Assuming
USC - MATH - 218
Student t Distribution In 1908, William S. Gosset, a chemist and statistician at the Guinness brewery in Dublin, noticed that the usual statistical practice of his day introduced small errors when sample sizes are small. The standard practice was to
N. Georgia - CIV - 3020
L'Art Romantique(Premire Moiti du 19 Sicle)L'Art Romantique Compar l'Art ClassiqueLE NOCLASSICISME (18me Sicle)- Sujets emprunts l 'histoire grecque et romaine -Composition quilibre et calme -Une certaine simplicit -La Beaut pure Jacques-Lou
USC - MATH - 525
FINAL EXAM, MATH 525A DUE DECEMBER 16, 2002Directions. Do all of the following problems; they are weighted equally. Start each problem on a fresh sheet of paper, and put your name on each sheet. Return the exam to me (via the office staff is fine)
N. Georgia - FREN - 4190
L'Art Romantique(Premire Moiti du 19 Sicle)L'Art Romantique Compar l'Art ClassiqueLE NOCLASSICISME (18me Sicle)- Sujets emprunts l 'histoire grecque et romaine -Composition quilibre et calme -Une certaine simplicit -La Beaut pure Jacques-Lou
USC - MATH - 425
SOLUTIONS FOR SECOND MIDTERM, MATH 425A FRIDAY, NOVEMBER 21, 1997Problem 2. Prove: if an bn cn for all positive integers n, and the series n an and n cn are both summable, then the series n bn is also summable. Proof. The complication here is tha
USC - MATH - 218
FINAL EXAM SOLUTIONS MATH 218 JUNE 30, 1998This was essentially the common nal given in the Spring of 1998. (I cut out the 11th problem on the common nal.) Problem 1. A company with 1000 employees has the following distribution of salaries for thre
USC - MATH - 218
Final Grade Book for Math 218, Section 50334, Fall 2005The gradesheet found on the next two pages reflects the letter grades for the various components of the course. To find the raw scores, look nearby where you found this table in the website. Th
N. Georgia - BADM - 4655
Improvement Calls for New Ways of ThinkingThought ProvokersNOTE: Many of these are quotes or paraphrases can be attributed to W. Edwards Deming or Peter Senge. Where a direct quote, the name follows the statement. 1. Describing is easier than expla
USC - MATH - 218
Raw Scores for Math 218, Section 50334, Fall 2005Code biq bov buv cev cuh cyc cyd faq fic gik gyf gym hac hel hum hyd hyw joj lax lib luq myv nik nir nym pab pic pod pydMid1 80 89 45 65 90 74 86 88 77 89 90 79 72 92 81 45 82 90 89 71 69 93 63 83
USC - MATH - 218
Math 218, Fall 2005, 9:00-9:50MWF Class Number 50334RThis document is distributed to the class in printed form, but the most current is always available on the World Wide Web, at URL <http:/imperator.usc.edu/~bruck/classes/fall2005/math218.09/>Pro
N. Georgia - BADM - 4655
Comments Selected 10. 100% inspections does not guarantee 100% correct; 200% inspection is worse! (15 what) -2. They did it! The enemy is out there. (3 no; 7 what) -4. People act rational to the systems we create. (8 No) 17. Competition and Rewards d
N. Georgia - BADM - 4655
Change Concepts IdentifiedSource: Langley, Gerald J., Nolan, Kevin M., Nolan, Thomas W., Norman, Clifford L., and Provost, Lloyd P., The Improvement Guide: A Practical Approach to Enhancing Organizational Performance, Jossey-Bass, 1996Eliminate Was
N. Georgia - BADM - 4655
Change ConceptsThe following are quotes from Langley, Nolan, Nolan, Norman and Provost, The Improvement Guide: A Practical Approach to Enhancing Organizational Performance, San Francisco: JosseyBass, 1996. They appear in the appendix "A Resource Gui
N. Georgia - BADM - 4655
Pizza WorksYou work at a company that makes pizzas. There are two kinds of pizza: Regular and Deluxe. You make $.50 profit on each Regular pizza and $.75 on each Deluxe. Business is good; you can sell all the pizzas you can make. You have already co
N. Georgia - BADM - 4655
Linear ProgrammingFormulating the linear programming problem: Read the problem. Determine the decision variables Break the problem down in English What is the objective? What limits your choices? (What are the constraints?) Would negative values for
N. Georgia - BADM - 4655
N. Georgia - BADM - 4655
N. Georgia - BADM - 4655
LP Special cases 1. max x1 x2 s.t. x1 + x2 5 x1 + 2x2 4 x1, x2 02.max x1 + x2 s.t. -x1 + x2 1 x1 2x2 2 x1, x2 0 max 2x1 + 2x2 s.t. x1 + x2 5 -x1 + x2 2 3x1 + x2 12 x1, x2 0 max 7x1 + 5x2 s.t. 7x1 + 3x2 42 2x1 3x2 6 9x1 + 13x2 = 117
N. Georgia - BADM - 4655
Using Excel to Solve Linear Programming ProblemsNote: The Pizza Problem will be must to illustrate the use of Excel. Step 1: Formulate the linear programming problem. Define the decision variables o Example: Reg = the number of regular pizzas to
N. Georgia - BADM - 4655
What-ifs for formulation problems 1. Captain Johns a) Solve using Excel b) What if a new captain appeared that was willing to work for $50 per trip? c) How much should he be willing to pay an extra crew member? d) What if he was able to rent another
N. Georgia - BADM - 4655
A local Quick Stop Food Store (1 check out) advertises fast service and short lines. To confirm their commitment to this advertisement, they have the following policies: 1. 2. 3. 4. The average number of people in line will not exceed 3 The likelihoo
USC - MATH - 218
Code bax caq cas cyn dyc fet fim gon gyv haj hen hyw jex jym kev kop kuh leh loz mif myj nan nus peh pod pul puz pyd ren rez ryq sap saq sid teb tev tod vat waq wib wyt zar ziq zuxCourse BB+ AAAIN A AB AB D+ A C+ UW B+ B C+ BD+ BB+ B C+ C+ A B+ C+
USC - MATH - 218
Code bax caq cas cyn dyc fet fim gon gyv haj hen hyw jex jym kev kop kuh leh loz mif myj nan nus peh pod pul puz pyd ren rez ryq sap saq sid teb tev tod vat waq wib wytMid1 Mid2 88 64 90 76 97 86 87 87 92 91 100 95 93 96 95 90 54 98 85 82 83 61 39
N. Georgia - BADM - 4668
Dvorak Simplified Keyboard In the 1920s or 1930s the Dvorak Simplified Keyboard (shown below) was introduced as replacement for the QWERTY (standard keyboard). This keyboard has the potential to increase operator efficiency by over 40%.Using the sa
USC - MATH - 218
Math 218, Spring 2005, 11:00-11:50MWF Class Number 50316RThis document is distributed to the class in printed form, but the most current is always available on the World Wide Web, at URL <http:/imperator.usc.edu/~bruck/classes/spring2005/math218.11/
N. Georgia - BADM - 3120
Solving Simultaneous Equations Using Row Reduction and a Matrix Approach Example Step 1: Identify the system of equations x+y+z=6 xy+z=2 3x + y z = 2 x 1 1 3 y 1 -1 1 z 1 1 -1 b 6 2 2Step 2: Write in matrix notation (Create one column for each var
USC - MATH - 125
First Mathematica Assignment Math 125, Class #50304 Due Thursday, September 9, 1999 Before attempting these problems, you should get a copy of the Mathematica Crib Sheet, available from <http:/mathlab.usc.edu/help/mmacrib.pdf> (you will need to have
N. Georgia - BADM - 4668
MGMT 4668 Fall 2005 Seminar in Management 3 semester hours Dr. Kim Melton Dahlonega: 132 Newton Oakes Center Gainesville: 142 Dunlap/Mathis (Academic I) kmelton@ngcsu.edu(706) 867-2724 (770) 718-3681Office Hours 8:30 9:30 Tuesday and Thursday i
USC - MATH - 425
First Assignment Math 425a, Fall 2003 Due Tuesday, September 2, 2003 Directions This is a mini-development of a piece of the development of the natural numbers via Peano's axioms. Problem 1. Prove: for all natural numbers m, n: if m + n = 0, then m =
USC - MATH - 425
THIRD HOMEWORK ASSIGNMENT MATH 425A, FALL 2003 DUE TUESDAY, SEPTEMBER 16, 2003Do problems 1012 of Chapter 1, and also problem 17. Problem 1. (Warm-up, easy) Let a, b, c, d be real numbers with a < b and c < d. Construct bijections [a, b] [c,
N. Georgia - BADM - 4668
Potential Articles for MGMT 4668 - Fall 2005 # 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 Title 7 Transformations of Leadership A Conversation on Leadership A Health Care Agenda for Business Act
USC - MATH - 425
FOURTH HOMEWORK ASSIGNMENT MATH 425A, FALL 2003 DUE TUESDAY, SEPTEMBER 30, 2003Do problems 48, 11, 20 of Chapter 2. Problem 1. A relation R on a set S is said to be a partial order if (a) R is reflexive: x R x for all x S, (b) R is anti-symmetric:
USC - MATH - 425
FIFTH HOMEWORK ASSIGNMENT MATH 425A, FALL 2003 DUE TUESDAY, OCTOBER 6, 2003Problem 1. Suppose (M, d) is a metric space, and put d(x, y) 1 + d(x, y) for x, y M . Prove that is also a metric on M , and that the d-open sets are exactly the -open set
USC - MATH - 425
10th Assignment Math 425a, Fall 2003 Due Tuesday, November 25, 2003 Problem 1. In what follows, you may assume that x < log(1 + x) < x for x > 0. x+1 (a) Show that the sequencenan =k=11 - log n kis decreasing. (b) Show that the sequencenbn