1) The lecture covered Bayes' theorem, random variables, probability distributions including binomial, Poisson, and exponential distributions.
2) The binomial distribution describes the probability of successes in independent Bernoulli trials while the Poisson distribution describes independent arrivals in a fixed time period.
3) Key characteristics of these distributions are presented such as the formulas for mean and variance. Examples are provided to illustrate real world applications.
1) The lecture covered Bayes' theorem, random variables, probability distributions including binomial, Poisson, and exponential distributions.
2) The binomial distribution describes the probability of successes in independent Bernoulli trials while the Poisson distribution describes independent arrivals in a fixed time period.
3) Key characteristics of these distributions are presented such as the formulas for mean and variance. Examples are provided to illustrate real world applications.
• Bayes’ Theorem • The binomial distribution is used to find the probability of a ▪ Update prior probability assessment based on new information specific number of successes out of 𝑛 independent trials of a ➢ General form Bernoulli process. ➢ Venn diagram • Random variables and probability distributions • 𝑛 = the number of trials ▪ A mapping from all possible outcomes to real numbers. • 𝑝 = the probability of success on any single trial ▪ Discrete → histogram, cumulative probability graph • 𝑟 = the number of successes ▪ Continuous → cumulative distribution function (CDF), probability density function (PDF) • ! means factorial (0! = 1) 𝑛! ➢ The probability value of each outcome of the continuous RV is 0. • 𝑃 𝑟; 𝑛, 𝑝 = ∙ 𝑝𝑟 ∙ 1 − 𝑝 𝑛−𝑟 ➢ In the graph of PDF, the area underneath the curve represents probability. 𝑟! 𝑛−𝑟 ! 5! ▪ P (Getting 4 heads in 5 tosses) = ? ∙ 0.54 ∙ 1 − 0.5 5−4 = 0.15625 ▪ Summary statistics → expected value and variance 4! 5−4 ! • Binomial distribution & Bernoulli process • If 𝑋~𝐵 𝑛, 𝑝 , then 𝐸 𝑋 = 𝑛𝑝 and 𝑉 𝑋 = 𝑛𝑝 1 − 𝑝 . 3 Binomial Distribution • The probability of obtaining specific outcomes in a Bernoulli Chapter 01 Probability Concepts process is described by the binomial probability distribution. and Applications • Characteristics of Bernoulli process ▪ Each trial in a Bernoulli process has only two possible outcomes (e.g., IIMT3636 success vs. failure, yes vs. no, head vs. tail, etc.). Faculty of Business and Economics ▪ The probability stays the same from one trial to the next. ▪ The trials are statistically independent. The University of Hong Kong ▪ The number of trials is a positive integer. Instructor: Dr. Yipu DENG • Example ▪ Tossing a coin for several times and counting the number of heads 6 8 Poisson Distribution Poisson Distribution in Excel • A Poisson RV describes the number of independent arrivals during a unit period of time. ▪ Patients arriving at a health clinic ▪ Customers arriving at a bank window ▪ Passengers arriving at an airport • 𝜆 = average number of arrivals per unit of time • 𝑒 = 2.718, the base of the natural logarithm • 𝑥 = the actual number of arrivals (0, 1, 2, …) 𝑒 −𝜆 ∙𝜆𝑥 • 𝑃 𝑥; 𝜆 = (Note: if x’ time length ≠ a unit period of time, adjust 𝜆) 𝑥! ▪ If the average number of arrivals per hour is 𝜆, the average number of arrivals in every two hours should be 2𝜆. • If 𝑋~𝑃𝑜𝑖𝑠𝑠𝑜𝑛 𝜆 , then 𝐸 𝑋 = 𝜆 and 𝑉 𝑋 = 𝜆. 5 7 Binomial Distribution in Excel Poisson Distribution 𝑃 𝑋≥1 𝑒 −𝜆 ∙ 𝜆𝑥 = 1−𝑃 𝑋 =0 𝑃 𝑥; 𝜆 = • Example = 1 − 𝑒 −0.13 = 12.2% 𝑥! ▪ Assume the number of financial crisis in one year follows Poisson distribution. ▪ How many crises are there in 100 years? ▪ What is the probability of seeing at least one crisis next year? 10 12 Exponential Distribution Exponential Distribution in Excel • The exponential distribution often describes the time required to serve a customer or the interarrival time or the lifespan of a product (e.g., a light bulb). → continuous • If 𝑋~𝐸𝑋𝑃 𝜆 , then 𝑓 𝑥 = 𝜆𝑒 −𝜆𝑥 and 𝐹 𝑥 = 𝑃 𝑋 ≤ 𝑥 = 1 − 𝑒 −𝜆𝑥 , where 𝜆 = 1/𝐸 𝑋 . ▪ 𝜆 = average number of arrivals per unit of time ▪ E(X) = average service time or interarrival time or lifespan ➢ Example: a chef needs 0.5 hours on average to cook one meal. X = the actual waiting time between completing two meals. E(X) = 0.5, which means the average waiting time between completing two meals. (Note: The unit used in describing 𝜆 determines the unit for the time X.) 𝜆 = 2, which means the average rate at which meals are cooked per hour. • 𝐸 𝑋 = 1/𝜆 and 𝑉 𝑋 = 1/𝜆2 . 9 11 Binomial and Poisson Distributions Exponential Distribution • It turns out the Poisson distribution is just a special case of the • Example 𝐹 𝑥 = 𝑃 𝑋 ≤ 𝑥 = 1 − 𝑒 −𝜆𝑥 binomial distribution — where the number of trials is very large, ▪ The service person can install new doors at a rate about 3 per and the probability of success in any given one is small. hour. • We can use a Binomial distribution with parameters (𝑛, 𝑝) to ▪ The service time is exponentially distributed. approximate a Poisson distribution with rate 𝜆 by setting 𝑛 to a large number and 𝑝 = 𝜆/𝑛. ▪ What is the probability that the time to install a new door will be 30 minutes or less? ▪ If n≥100 and np≤10, the approximation can be very good. Poisson, l = 5 0.2 𝑃 𝑋 ≤ 0.5 0.15 = 1 − 𝑒 −3∗0.5 = 1 − 𝑒 −1.5 = 77.7% 0.1 0.05 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 14 16 Exponential and Poisson Distributions Uniform Distribution in Excel • Example • The 𝑟𝑎𝑛𝑑() function generates a number from a uniformly ▪ Suppose customer arrivals at a retail store per hour follow distributed random variable on [0,1] Poisson with mean 𝜆 = 3. Given that a customer arrived at 2:30pm, what is the probability that the next customer will arrive • For a general uniform distribution on [a,b], we can use before 3:00pm? (Hint: how to define the random variable?) “= 𝑎 + 𝑟𝑎𝑛𝑑() ∗ 𝑏 − 𝑎 ” ▪ 𝑃𝑃 𝑋 ≥ 1; 𝜆 = 1.5 or 𝑃𝐸 𝑌 ≤ 0.5; 𝜆 = 3 13 15 Exponential and Poisson Distributions Uniform Distribution • Exponential and Poisson are related: if the number of • A random variable 𝑋 is uniformly distributed if it takes arrivals per time period follows Poisson(𝜆0 ), then the any value equally likely from a finite interval 𝑎, 𝑏 . time between arrivals (1/ 𝜆1 ) follows exponential Probability density function Cumulative distribution function distribution EXP(𝜆1 ) with an identical rate 𝜆1 = 𝜆0 . 𝑥−𝑎 = 𝑏−𝑎 ▪ Example: if the number of phone calls arriving at a customer service center follows a Poisson distribution with a mean of 10 calls per hour (𝜆 = 10), the time between each phone call would be exponentially distributed, with the average time between calls as 1/10 hours or 6 minutes. 1 1 • 𝐸 𝑋 = 𝑎+𝑏 ;𝑉 𝑋 = (𝑏 − 𝑎)2 2 12 18 20 Normal Distribution (PDF) Normal Distribution • Normal distribution with different 𝜇. (shape unchanged) | | | 40 m = 50 60 Smaller m, same s | | | m = 40 50 60 Larger m, same s | | | 40 50 m = 60 17 19 Normal Distribution Normal Distribution • The normal distribution is the most popular and useful • Normal distribution with different 𝜎. (always symmetric) continuous probability distribution. ▪ E.g., return of a stock portfolio, forecast errors, and test scores. • The probability density function is 1 𝑥−𝜇 2 Same m, smaller s − 𝑓 𝑥 = ∙ 𝑒 2𝜎2 𝜎 2𝜋 Same m, larger s • The normal distribution is specified completely when we know the mean 𝐸(𝑥), 𝜇, and the standard deviation, 𝜎. We often use the notation 𝑁 𝜇, 𝜎 2 . m 22 24 Standard Normal Distribution: Example In-class Exercise • If 𝑋 follows normal distribution with mean 𝜇 = 100, 𝜎 = • If 𝑋 follows normal distribution with mean 𝜇 = 100 and 15, and we are interested in finding the probability that X s.d. 𝜎 = 22, then 𝑃 𝑋 < 120 = ? is less than 130, then z = ? ▪ Hint: P(X ≤ x) = P (Z ≤ z) 𝑋−𝜇 120−𝜇 • 𝑃 𝑋 < 120 = 𝑃 𝑋 − 𝜇 < 120 − 𝜇 = 𝑃 < 𝜎 𝜎 𝑋−𝜇 𝑥−𝜇 • Define 𝑍 = . 𝑧= 𝜎 𝜎 130−100 • 𝑃 𝑋 < 120 = 𝑃 𝑍 < 0.91 = 0.8186 = =2 15 21 23 P(Z<-z) = P(Z>z) Standard Normal Distribution Standard Normal Distribution: Z Table P(Z<0.57)? 0.7157 If P(Z<z) = 0.65, z? 0.385 z = 2.00, the area is 0.97725, P(X<130)? 0.97725 Areas under the normal curve 𝑋−𝜇 If 𝑿~𝑵 𝝁, 𝝈𝟐 , then 𝑍 = 𝜎 follows standard normal distribution. Here, Z represents the number of standard deviations from X to the mean, 𝜇 26 28 Haynes Construction Company Haynes Construction Company • Builds three- and four-unit apartment buildings • Suppose Haynes can hire more workers to reduce the • Total construction time follows a normal distribution mean construction time while keeping the standard • For triplexes, μ = 100 days and σ = 20 days deviation constant (i.e., 20 days). Hence, to ensure a penalty is avoided with probability 0.95, what should be • Contract calls for completion in 125 days the new mean? • Late completion will incur a severe penalty fee • Probability of completing in 125 days? 𝑋 ′ −𝜇 • Objective: find 𝜇 such that 𝑃 𝑋′ < 125 = 𝑃 ቀ < 20 125−𝜇 ቁ = 0.95 20 𝑃 𝑋 < 125 = 𝑃 𝑍 < 1.25 = 0.8944 • Step 1: 𝑧0.95 = 1.645 125−𝜇 m = 100 days • Step 2: find 𝜇 such that = 1.645; 𝜇 = 92.1 s = 20 days 20 x = 125 days 25 27 Finding the x value Haynes Construction Company • If 𝑋 follows normal distribution with mean 𝜇 = 100 and s.d. • If finished in 75 days or less, bonus = $5,000. Probability of bonus? 𝜎 = 22 and we want 𝑋 to be less than a certain value with • 𝑃 𝑋 < 75 = 𝑃 𝑍 < −1.25 probability 0.8, then how to find this value? Or 𝑃 𝑋 < ? = • = 𝑃 𝑍 > 1.25 = 1 − 𝑃 𝑍 < 1.25 = 1 - 0.8944 = 0.1056 0.8. • What is the probability of completing between 110 and 125 days? • Step 1: Finding the z value corresponding to 𝑃 𝑋 < 𝑥 . • 𝑃 110 < 𝑋 < 125 = 𝑃 𝑋 < 125 − 𝑃 𝑋 < 110 ▪ If we need 𝑥 for 𝑃 𝑋 < 𝑥 , find 𝑧 corresponding to 𝑃 𝑋 < 𝑥 . 𝑧0.8 = 0.84 = 0.8944 – P(Z<0.5) If we need 𝑥 for 𝑃 𝑋 > 𝑥 , find 𝑧 corresponding to 1 − 𝑃 𝑋 > 𝑥 . = 0.8944 – 0.6915 • Step 2: Setting 𝑥 to 𝑥𝑧 = 𝜇 + 𝑧 ∙ 𝜎. = 0.2029 𝑋−𝜇 ▪ 𝑃 𝑋 < 𝑥𝑧 = 𝑃 𝑋 < 𝜇 + 𝑧 ∙ 𝜎 = 𝑃 𝜎 < 𝑧 = 𝑃 𝑍 < 𝑧 = 𝑃 𝑋 < 𝑥 𝑃 𝑋 < 100 + 𝑧0.8 × 22 = 𝑃 𝑋 < 118.48 = 0.8 110 days 125 days 31 33 In-Class Exercise In-Class Exercise • When a man passes the airport security check, they • A candidate for public office has claimed that 60% of discover a bomb in his bag. He explains. “Statisticians voters will vote for her. If 5 registered voters were show that the probability of one bomb being on an sampled, what is the probability that exactly 3 would say airplane is 1/10,000. However, the chance that there are they favor this candidate? (hint: utilize BINOM.DIST in two bombs on one plane is 1/10,000,000. So, I am much Excel) safer …” • Suppose the statisticians are right and it is impossible to Binomial distribution with n = 5, p = 0.6, r = 3, prob = 0.3456 have more than two bombs on an airplane. Do you agree with the man? No. The conditional probability of flying with another bomb is 1/1,000. 29 32 Normal Distribution in Excel In-Class Exercise • A class contains 30 students. Ten are female (F) and U.S. citizens (U); 12 are male (M) and U.S. citizens; 6 are female and non-U.S. citizens (N); 2 are male and non- U.S. citizens. • A name is randomly selected from the class roster, and it is female. What is the probability that the student is a U.S. citizen? P(U|F) = P(UF)/P(F) = (10/30)/[(10+6)/30] = 0.625 2 4 Decision makers have The Science of Decision Making Thompson Lumber Company no control over it. • Making a decision is about choosing among alternatives. • List all possible alternatives (i.e., course of action or strategies) ▫ Whether to pursue a graduate study (BA or Marketing); whether to • Identify possible outcomes/states of nature (e.g., high vs. low demand) seek a job ▫ Whether to buy a stock and how much money to invest (e.g., all, half, • List the payoff of the combination of alternatives and outcomes 1/3); other investment opportunities • Select and apply the decision theory model, make decisions ▫ Whether to expand the product line and how to expand (e.g., new or existing technology) STATE OF NATURE • Six Steps in Decision Making FAVORABLE MARKET UNFAVORABLE MARKET ▫ Clearly define the problem (Goal to achieve) ALTERNATIVE (profit in $) (profit in $) ▫ List the possible alternatives Construct a large plant 200,000 –180,000 ▫ Identify the possible outcomes or states of nature ▫ List the payoff of each alternative in each state of nature Construct a small plant 100,000 –20,000 ▫ Select one of the decision theory models Do nothing 0 0 ▫ Apply the model and make your decision Decision/payoff table 3 Thompson Lumber Company • John Thompson, the founder and president, needs to make the decision on whether to expand his product line by manufacturing Chapter 02 and marketing a new product. • Define the problem Decision Analysis à whether to expand the product line IIMT3636 Faculty of Business and Economics The University of Hong Kong Instructor: Dr. Yipu DENG 6 8 Decision Making under Risk Decision Making under Risk • EMV is the weighed sum of possible payoffs for each alternative. • EVwPI = ∑ best payoff in state - . probability of state - STATE OF NATURE Expected Monetary Expected FAVORABLE UNFAVORABLE Value STATE OF NATURE Monetary ALTERNATIVE MARKET (profit in $) MARKET (profit in $) (EMV, $) FAVORABLE UNFAVORABLE Value ALTERNATIVE MARKET (profit in $) MARKET (profit in $) (EMV, $) Construct a large plant 200,000 –180,000 –9,000 Construct a small plant 100,000 –20,000 34,000 Construct a large plant 200,000 –180,000 –9,000 Do nothing 0 0 0 Construct a small plant 100,000 –20,000 34,000 Probability 0.45 0.55 Do nothing 0 0 0 Best payoff 200,000 0 90,000 Probability 0.45 0.55 EVwPI If John maximizes the EMV, then he should choose to construct a small plant. EVPI = EVwPI – Best EMV = $90,000 – $34,000 = $56,000 < $65,000. Don’t buy! 5 7 Types of Decision-Making Environments Decision Making under Risk • Decision making under certainty • Scientific Marketing, Inc. offers analysis that will provide ▫ The decision maker knows with certainty the consequences of every alternative or decision choice. certainty about market conditions. S.M. would charge ▫ Example: have $1,000 to invest for 1 year (open a savings account $65,000 for the information. Should John buy the with 4% interest vs. invest in a government bond with 6% interest) • Decision making under risk information? ▫ There are several possible outcomes for each alternative, and decision maker knows the probabilities of each outcome. ▫ Example: get $10 if we roll a 5 on a standard dice à maximize • To make this decision, John has to evaluate the expected expected monetary value or minimize opportunity loss value of perfect information (EVPI) by computing the • Decision making under uncertainty expected value with perfect information (EVwPI) and ▫ There are several possible outcomes for each alternative, and decision maker does not know the probabilities of the various the best EMV under risk. outcomes. ▫ Example: the probability of success of a new product 10 12 Decision Making under Risk In-class Exercise: Café du Donut Favorable Market VwPI Unfavorable Market • Monetary Payoff (Profit) Table OL2 OL3 VwPI • Cost = $40, revenue = $60 MV1 OL2 MV2 MV3 OL1 MV3 MV2 D=4 D=5 D=6 D=7 D=8 D=9 D = 10 EMV MV1 4x60 6x40 6x60 6 x40 Q=6 0 60Go 120 120 120 120 120 105 - - Option Option Option Option Option Option O 120 120 10 120 120 105 1 2 3 1 2 3 60 7 x 40 4-40 7x 60 7 x 40 Q=7 x 202 O 80 140 140 140 140 104 - - - 40 80 140 140 140 140 104 Prob. 0.05 0.15 0.15 0.20 0.25 0.10 0.10 EVwPI EMVLQ = 6) = 0x0 05 + 60 x 0 . 15 + 120 x (0 . 15 +0 . 20 + 0 . 25 +0 . 10 + 0 - 10) = 105 EOL2 . EOL1 EOL3 • Should we reduce the order size from 6 to 5? What is the EMV of Q=5? EMV1 EMV2 EMV3 Option 1 Option 2 Option 3 • If we can only choose between 6 and 7, what is the EVPI? 9 11 Decision Making under Risk In-class Exercise: Café du Donut • To minimize expected opportunity loss (EOL) • The Café buys donuts each day for $40 per carton of 20 dozen • Opportunity loss = optimal payoff – actual payoff (in a given state of nature) donuts. Any cartons not sold are thrown away at the end of STATE OF NATURE Expected the day. If a carton is sold, the total revenue is $60. Monetary FAVORABLE FAVORABLE Mkt UNFAVORABLE UNFAVORABLE Mkt EOL Value DAILY DEMAND PROBABILITY CUMULATIVE ALTERNATIVE (Opp. Loss MARKET in $) (profit in $) (Opp. Loss MARKET in $) (profit in $) (EMV, (in $) (CARTONS) PROBABILITY Construct a large plant 0 200,000 180,000 –180,000 99,000 –9,000 • The original plan 4 0.05 0.05 Construct a small plant 100,000 20,000 –20,000 56,000 34,000 is to order 6 5 0.15 0.2 Do nothing 200,000 0 0 90,000 0 cartons per day. 6 0.15 0.35 Probability 0.45 0.55 Should the Café 7 0.20 0.55 • Minimizing EOL always results in the same decision as maximizing increase the 8 0.25 0.8 9 0.10 0.9 EMV. order size to 7? • The minimum EOL always equal the EVPI. 10 0.10 1.0 Total 1.00