1. Binomial Distribution

Definition — Bernoulli Process

  1. The experiment consists of repeated trials.
  2. Each trial is a success or failure.
  3. is the same on every trial ().
  4. Trials are independent.

Definition — Binomial Distribution

= number of successes in Bernoulli trials:

Why: a specific sequence with successes has probability (independence), and there are positions for the successes.

Theorem — Mean and Variance

Proof. Write where if trial is a success. and . The are independent, so and .

Worked Example

20% of chips from a supplier are out of spec. In a sample of :

  • , ,
Binomial(n=10, p=0.2); shaded: P(X ≤ 2) = 0.6778 0.107 0.268 0.302 0.201 0.088 0.026 0 1 2 3 4 5 6 7 8

⚠️ “At least” / “more than” → use with the correct cut-off: .


2. Multinomial Distribution

Definition — Multinomial Distribution

Each of independent trials results in one of outcomes with probabilities (). If counts outcome :

with . Binomial is the case .

Worked Example

Packets are routed via links A, B, C with probabilities 0.5, 0.3, 0.2. Of 6 packets, probability that 3 go via A, 2 via B, 1 via C:


3. Hypergeometric Distribution

Definition — Hypergeometric Distribution

Sampling items without replacement from items of which are “successes”:

Theorem — Mean and Variance


The factor is the finite population correction.

Worked Example — Acceptance Sampling

A lot of parts contains defectives. Inspect :

  • ,

Theorem — Binomial Approximation

If is small relative to (rule of thumb ), .

⚠️ With replacement (or an infinite process) → binomial. Without replacement from a small finite lot → hypergeometric.


4. Geometric and Negative Binomial (Pascal) Distributions

Definition — Negative Binomial Distribution

= trial number on which the -th success occurs:

Why: the -th trial must be a success, and the first trials must contain exactly successes.

Definition — Geometric Distribution ()

= trial number of the first success:

Theorem — Moments

MeanVariance
Geometric
Negative binomial

Proof (geometric mean). .

The negative binomial is a sum of independent geometric waiting times, which gives its moments.

Theorem — Memoryless Property (Geometric)


Proof. (first trials all failures), so the ratio is .

Worked Examples

  • A connection attempt succeeds with . First success on the 5th try: . Needing more than 5 tries: . Expected tries: .
  • With , probability that the 3rd success occurs on the 7th trial: .

⚠️ Some books define the geometric as the number of failures before the first success (, mean ). Check which convention your course uses.


5. Poisson Distribution and the Poisson Process

Definition — Poisson Process

Events occurring in time (or space) such that:

  1. Counts in disjoint intervals are independent (no memory).
  2. The probability of one event in a short interval of length is , proportional to its length.
  3. The probability of two or more events in a very short interval is negligible.

Definition — Poisson Distribution

The number of events in an interval of length is Poisson with mean :

Theorem — Mean and Variance

Proof (mean).

The same trick on gives .

Worked Examples

  • A server receives on average 3 requests per second ():
  • Errors arrive at per hour. In a 3-hour shift, and .
Poisson(μ=3); shaded: P(X ≤ 2) = 0.4232 0.050 0.149 0.224 0.224 0.168 0.101 0.050 0.022 0 1 2 3 4 5 6 7 8 9 10

⚠️ Always scale the rate to the interval asked about: (rate × length), with matching units.


6. Binomial–Poisson Approximation

Theorem — Poisson Limit of the Binomial

If and with fixed, then

Proof sketch. With :

Rule of thumb: good when and (or , ).

Worked Example

A batch of solder joints, each defective with . :

MethodResult
Exact binomial0.85730
Poisson, : 0.85712

Summary Table

DistributionModelspmfMeanVariance
Binomial# successes in trials
Hypergeometric# successes, no replacement
Geometrictrial of 1st success
Negative binomialtrial of -th success
Poisson# events in interval