Business Statistics – Discrete Distributions (MDB2013 Chapter 4)
Teaching & Learning Outcomes
- Compute probabilities from various discrete distributions.
- Utilize discrete probability plots to determine financial applications.
- Calculate probabilities specifically for the Binomial, Poisson, and Hypergeometric distributions.
Probability Distribution – Definition & Characteristics
- Definition: A listing of all experiment outcomes with their associated probabilities.
- Key characteristics:
- Probabilities lie in the interval [0,1].
- Outcomes are mutually exclusive.
- Outcomes are exhaustive → \sum P(x)=1.
Illustrative Example: 3 Coin Tosses
- Experiment: Count heads when a coin is tossed 3 times.
- Possible values of X: 0,1,2,3 heads.
- Probabilities:
- P(0)=\frac{1}{8}=0.125
- P(1)=\frac{3}{8}=0.375
- P(2)=\frac{3}{8}=0.375
- P(3)=\frac{1}{8}=0.125
- Check: 0.125+0.375+0.375+0.125=1.0 (distribution validated).
Random Variables
- Outcome values produced by a chance experiment.
- Examples:
- Number of employees absent on Monday (quantitative discrete).
- Grade level of team members (qualitative categorical variable but treated as a random variable).
Types
- Discrete: Assumes clearly separated values (e.g., number of credit cards a customer holds).
- Continuous: Infinite possible values in an interval (e.g., flight time KLIA–Langkawi; snowfall depth).
Mean, Variance & Standard Deviation of a Discrete Distribution
- Mean (Expected value): \mu = \sum xP(x).
- Variance: \sigma^{2}= \sum (x-\mu)^{2}P(x).
- Standard deviation: \sigma = \sqrt{\sigma^{2}}.
Example: Johan’s Car Sales on Saturday
- Distribution (condensed): P(0)=0.1,\ P(1)=0.2,\ P(2)=0.3,\ P(3)=0.3,\ P(4)=0.1.
- Calculations:
- \mu = 2.1 cars.
- Tabled variance computation (\sum (x-\mu)^2 P(x)) → \sigma^{2}=1.29.
- Interpretation: On a typical Saturday Johan expects about 2.1 cars sold with moderate variability (SD \approx1.136).
Binomial Distribution
- Definition: Widely occurring discrete distribution meeting 4 criteria:
- Only two outcomes (success/failure).
- Fixed, known number of trials n.
- Constant success probability \pi across trials.
- Trials are independent.
- Probability mass function (PMF):
P(X=x)=C^{n}_{x}\pi^{x}(1-\pi)^{n-x} for x=0,1,\dots,n. - Mean & Variance:
- \mu = n\pi
- \sigma^{2}=n\pi(1-\pi)
Debit-Card Coffee-Shop Example (Zus Coffee, n=5,\pi=0.28)
- (a) Exactly one debit-card purchase:
P(1)=C^{5}_{1}(0.28)^{1}(0.72)^{4}. - (b) Full distribution: Evaluate PMF for x=0\rightarrow5.
- (c) P(X\ge 3)=P(3)+P(4)+P(5).
- (d) P(X\le 2)=1-P(X\ge3) (or directly sum).
- (e) \mu =5(0.28)=1.4, \sigma^{2}=5(0.28)(0.72)=1.008.
- Verify all 4 binomial requirements (yes – meets each).
Binomial Tables Example (Dropped Calls, n=6,\pi=0.05)
- Use Table 6-2.
- (a) P(0)=0.7351 (value from table).
- (b) P(1)=0.2326, etc. through P(6)=1.6\times10^{-8}.
Hypergeometric Distribution
- Used when sampling without replacement from a finite population where n/N>0.05 (lack of independence).
- Conditions:
- Two outcome categories.
- Fixed sample size n.
- Trials are dependent.
- Population finite and sampling w/out replacement.
- PMF:
P(X=x)=\dfrac{C^{S}{x}\,C^{N-S}{n-x}}{C^{N}_{n}}.
- N = population size, S = total successes in population.
Classroom Committee Example
- N=50,\ S=40,\ n=5,\ x=4.
- Probability:
P(4)=\dfrac{C^{40}{4}\,C^{10}{1}}{C^{50}_{5}} (compute via combinations).
Poisson Distribution
- Describes number of occurrences in a specified interval (time, distance, area, volume).
- Assumptions:
- Probability proportional to interval length.
- Intervals are independent.
- PMF (parameter \lambda = mean number of events per interval):
P(X=x)=\dfrac{\lambda^{x}e^{-\lambda}}{x!}. - Relation to Binomial: Limiting case when n\to\infty, \pi\to0, n\pi=\lambda.
Lost-Bags Example (Amal by Malaysia Airlines)
- Sample: 500 flights, 20 lost bags → overall rate \lambda_{500}=20.
- (a) Valid Poisson: counting occurrence per non-overlapping flight, independence presumed.
- (b) Mean per flight: \lambda =20/500=0.04 bags.
- (c) No bags lost on a flight: P(0)=e^{-0.04}=0.9608.
- (d) At least one lost: 1-P(0)=0.0392.
Poisson Table Example (Truck Breakdowns)
- Mean breakdowns per run: \lambda=0.30.
- From table:
- P(0)=0.7408
- P(1)=0.2222 (matches \lambda e^{-\lambda} check).
Conceptual Connections & Applications
- Financial modeling often uses binomial trees for option pricing; discrete plots help visualize risk.
- Quality-control auditing (hypergeometric) models defect sampling without replacement.
- Operations management (Poisson) forecasts arrival rates—e.g., call centers, traffic, insurance claims.
- Ethical note: Accurate statistical modeling prevents misallocation of resources, fosters fair decisions.
- Discrete Mean: \mu=\sum xP(x)
- Discrete Variance: \sigma^{2}=\sum (x-\mu)^{2}P(x)
- Binomial PMF: P(X=x)=C^{n}_{x}\pi^{x}(1-\pi)^{n-x}
- Binomial Mean/Var: \mu=n\pi,\ \sigma^{2}=n\pi(1-\pi)
- Hypergeometric PMF: P(X=x)=\dfrac{C^{S}{x}C^{N-S}{n-x}}{C^{N}_{n}}
- Poisson PMF: P(X=x)=\dfrac{\lambda^{x}e^{-\lambda}}{x!}
Study Tips
- Always check distribution requirements before applying a formula.
- For small n without replacement, prefer hypergeometric; for large n with small \pi, Poisson may approximate Binomial.
- Use tables or software to avoid factorial/combinatorial arithmetic mistakes.
- Draw probability plots to see skewness and assess variance vs. mean relationships.