Chapter 4: Counting Methods Counting methods are everywhere Product Rule Consider a situation where order matters, so that (๐๐๐๐๐๐ก! , ๐๐๐๐๐๐ก" ) is meaningfully different from (๐๐๐๐๐๐ก" , ๐๐๐๐๐๐ก! ). Example Company's retirement plan: 8 different bond funds and 12 different stock funds • For your portfolio: you are allowed to pick one bond fund and one stock fund How many possible ways are there to pick a portfolio? Product Rule Product rule for ordered pairs: If there are ๐ choices for the first action and ๐ choices for the second action, the number of possible choices for both actions is ๐๐. Generalized product rule: Suppose there are ๐ actions. If there are ๐! choices for the 1st , ๐" choices for the 2nd action, and so on through ๐# actions for the ๐-th action, the number of possible choices for the ๐ actions is given by the product of the numbers of choices, which is ๐! ๐" … ๐# . Product Rule Example Company's retirement plan: 8 different bond funds and 12 different stock funds • For your portfolio: you are allowed to pick one bond fund and one stock fund How many possible ways are there to pick a portfolio? = 8 x 12 = 96 Example Company's retirement plan: The retirement plan adds money-market fund options, so now they have 8 bond funds, 12 stock funds, and 4 money-market funds. How many ways are there to pick a portfolio with one bond fund, one stock fund, and one money-market fund? = 8 x 12 x 4 =384 Sum Rule Sum rule: If there are ๐ choices for the first action, ๐ choices for the second action, and only one of the actions can be taken, the number of possible choices is ๐ + ๐ . Example Retirement savings: If instead of choosing both a bond fund (8 choices) and a stock fund (12 choices), you are restricted to choose only one fund overall. What is the total number of possible choices? = 8+12 = 20 Generalized sum rule: Suppose there are ๐ actions. If there are ๐! choices for the 1st , ๐" choices for the 2nd action, and so on through ๐# actions for the ๐-th action, the number of possible choices when only one action can be chosen is given by the sum of the numbers of choices, which is ๐! + ๐" +…+ ๐# . Permutations and combinations Consider a group of n distinct objects. How many ways are there to choose a subset of size ๐ from the group? The first question to ask yourself: Does order matter or not? Example Board of Directors: Suppose there are 20 candidates for positions on a corporate board of directors. • How many ways are there to select a three-member board of directors (no titled positions)? Order doesn’t matter! • How many ways are there to select a three-member board of directors with titles of President, Vice President, and Treasurer specified? Order does matter! (Sue Pres, Joe VP, Kate Treas) Example Stock Portfolio: Suppose there are 100 possible stocks. • You want to invest equally in five stocks ( 20% in each). How many possible portfolios are there? Order doesn’t matter! • You want to form a 30%/25%/20%/15%/10% weighted portfolio of five stocks. How many possible portfolios are there? Order does matter! Permutations and combinations An ordered subset of distinct choices is called a permutation, and ๐$,# = # ๐๐ ๐๐๐๐๐ข๐ก๐๐ก๐๐๐๐ ๐๐ ๐ ๐๐ง๐ ๐ ๐กโ๐๐ก ๐๐๐ ๐๐ ๐๐๐๐๐๐ ๐๐๐๐ ๐ ๐๐๐๐๐๐ก๐ . An unordered subset of choices is called a combination, and ๐ถ$,# = # ๐๐ ๐๐๐๐๐๐๐๐ก๐๐๐๐ ๐๐ ๐ ๐๐ง๐ ๐ ๐กโ๐๐ก ๐๐๐ ๐๐ ๐๐๐๐๐๐ ๐๐๐๐ ๐ ๐๐๐๐๐๐ก๐ . ๐ถ!,# is often denoted ๐ and read as " ๐ choose ๐ ” ๐ General formula for permutations: ๐ factorial over (๐ − ๐) factorial ๐!,# = ๐! ๐−๐ ! Permutations Back to….Example Board of Directors: Suppose there are 20 candidates for positions on a corporate board of directors. • How many ways are there to select a three-member board of directors with titles of President, Vice President, and Treasurer specified? Need to determine the number of permutations of size 3 that can be formed from 20 objects. ๐$%,& = 20! 20! 20×19×18×17 … = = = 6840 20 − 3 ! 17! 17×16×15 … Example Stock Portfolio: Suppose there are 100 possible stocks. • You want to form a 30%/25%/20%/15%/10% weighted portfolio of five stocks. How many possible portfolios are there? ๐'%%,( = 100! = 100 99 98 97 96 = 9,034,502,400 95! Permutations in R Combinations Back to….Example Board of Directors: Suppose there are 20 candidates for positions on a corporate board of directors. How many ways are there to select a three-member board (with no titles)? Need to determine the number of combinations of size 3 that can be formed from 20 objects, or 20 ๐ถ$%,& or read as " 20 choose 3 ” 3 We know there are ๐$%,& permutations but notice that several of these permutations are the “same” in an unordered setting. Think about three candidates Jim, Kate, and Lucy, who would appear a total of six times within the ๐$%,& permutations: ( ๐ฝ, ๐พ, ๐ฟ ) , ( ๐ฝ, ๐ฟ, ๐พ ) , ( ๐พ, ๐ฝ, ๐ฟ ) , ( ๐พ, ๐ฟ, ๐ฝ ) , ( ๐ฟ, ๐ฝ, ๐พ ) , ( ๐ฟ, ๐พ, ๐ฝ ) Notice this is k factorial – 3!=3*2*1=6 Combinations General formula for combinations: ๐ถ$,# ๐! ๐ = = ๐ ๐ − ๐ ! ๐! Back to…How many ways are there to select a three-member board (with no titles)? ๐ถ"&,' 20! 20 ∗ 19 ∗ 18 20 = = = = 1,140 3 17! 3! 3∗2∗1 Permutation ๐$%,& Back to Example Stock Portfolio…How many ways are there to form an equally weighted five-stock portfolio ( 20% each)? ๐ถ'%%,( = 100! 100 ∗ 99 ∗ 98 ∗ 97 ∗ 96 100 = = = 75,287,520 5 95! 5! 5∗4∗3∗2∗1 Combinations in R Back to…How many ways are there to select a three-member board (with no titles)? Probabilities for equally likely choices In some situations, you might want to ask about the probability of seeing certain permutations or combinations In the case where any permutation or combination is equally likely, it is often possible to figure out the probability using math or using simulation methods. For now, let's focus on what's possible with math. Probabilities for equally likely choices Example Website Names Shorter website names are more valuable. In fact, it's highly unlikely that you would find a three-character website available for registration. 1. How many three-character websites are there? (By “character,” we mean any letter a through z or number 0 through 9 .) 26 letters (a –z) and 10 numbers 0-9 = 36. 36 for 1st char, 36 for 2nd char, 36 3rd char 36 36 36 = 36! 2. If one of these websites is chosen at random (each with the same chance), what is the probability that the website is composed entirely of alphabetic characters (no numbers)? # ๐๐ ๐๐๐ก๐ก๐๐๐ = 26 26 26 = 26! Probability = #$! !$! ≈0.377 3. If one of these websites is chosen at random (each with the same chance), what is the probability that the website has three distinct alphabetic characters? # of these is permutation ๐$),& = (26)(25)(24) Probability = (#$)(#&)(#') !$! ≈0.334 Probabilities for equally likely choices Back to Example Stock portfolio: Consider again the problem of forming a 5-stock equally weighted portfolio from 100 possible stocks, but now suppose that 20 stocks are classified as “tech” stocks and 80 stocks as “non-tech” stocks. How many equally weighted 5-stock portfolios have exactly 2 “tech” stocks and 3 “non-tech” stocks? Order of 2 T and 3 NT does not matter - combination Desired portfolio is of the form ( ๐, ๐, ๐๐, ๐๐, ๐๐ ) , where ๐ = ๐ก๐๐โ & ๐๐ = ๐๐๐ ๐ก๐๐โ 20 80 and for 3 non tech stocks # of ways: 2 3 20 80 Total number of portfolios is = 3 2 For 2 tech stocks # of ways: If you form a portfolio by picking five stocks (from the 100) at random, what is the probability that you have two or three “tech” stocks in the portfolio? Recall that the total number of possible portfolios is # with 2 T or 3 T = # with 2 T + # with 3 T = prob = $% $ 20 2 *% $% + & & '%% ( 100 , which is the denominator here 5 80 20 + 3 3 *% $ 80 2 Probabilities for equally likely choices Example The likelihood of “streaks”: Suppose that you flip a coin 100 times. What is the probability of having a streak of (at least) five consecutive heads within the 100 tosses? It's very difficult, though not impossible, to solve this problem using math. So, an alternative approach is to use computer simulations: Step 1: Simulate the experiment of flipping 100 coins. Step 2: Check whether or not there is a streak of five heads within the simulated 100 flips. Step 3: Repeat Steps 1 and 2 many times to determine the frequency (fraction) of times that a streak of five heads occurs within 100 flips. The following R code and output illustrates how to do it. The code conducts 100,000 simulations in order to approximate the desired probability. Example The likelihood of “streaks” R code: # of simulations # of tosses Frequency of 5 head streaks over 100,000 experiments Example The likelihood of “streaks” R code: Noise in estimating frequency 0.81156 Settles down