19.X. Poisson Distribution – Detailed Study

The Poisson distribution is a discrete probability distribution that gives the probability of a given number of events occurring in a fixed interval of time or space, given the events occur with a known constant mean rate and independently of the time since the last event. If 𝑋 is a Poisson random variable with mean 𝜆, then the probability mass function (PMF) is given by:

𝑃(𝑋= 𝑘) = \frac{e^{−𝜆}𝜆^{k}}{k!}, 𝑘 = 0,1,2,…

Where: • 𝜆 = average number of events in a given interval • 𝑒 ≈ 2.71828 • 𝑘! = factorial of 𝑘

The Poisson distribution models how many times an event happens in a specific timeframe, given that these events occur independently at a steady rate. The formula presented shows that to find the probability of observing exactly 'k' events, you need the average number of events (lambda) and the mathematical constant 'e'. The 'k!' part, the factorial of k, is essential for calculating probabilities when we look at multiple potential outcomes.

Imagine you're monitoring a streetlight to see how often it fails per month. If, on average, it fails 2 times a month, you can use the Poisson distribution to calculate the likelihood of it failing exactly 3 times in a given month.

1. Mean and Variance: Mean = 𝜆, Variance = 𝜆

2. Additive Property: If 𝑋 ∼ Poisson(𝜆₁) and 𝑌 ∼ Poisson(𝜆₂) are independent, then: 𝑋 + 𝑌 ∼ Poisson(𝜆₁ + 𝜆₂)

3. Memoryless Nature: Though primarily a property of the exponential distribution, the Poisson process assumes independent and memoryless events.

4. Skewness: Skewness = \frac{1}{\sqrt{𝜆}}

The properties of the Poisson distribution are crucial for understanding how it behaves. The mean and variance being equal (both 𝜆) implies that as you increase the average rate of events, the spread (or variability) around that average also increases. The additive property means that if you have two independent processes, their combined outcome also follows a Poisson distribution. Memoryless nature suggests that the future behavior of the process does not depend on the past, which is a unique characteristic in probability. Lastly, skewness shows how the symmetry of the distribution changes as the average number of occurrences grows.

Think of a bank receiving calls. If on average, they get 5 calls an hour (this is λ), then both the average and the variance of call arrivals for that hour are also 5. If one hour they get 3 calls and another hour 7 calls, it follows a similar pattern due to the Poisson properties.

The Poisson distribution is derived as a limiting case of the Binomial distribution when: • Number of trials 𝑛 →∞ • Probability of success 𝑝 → 0 • 𝑛𝑝 = 𝜆 remains constant

Let 𝑋 ∼ Binomial(𝑛,𝑝), then:

𝑃(𝑋 = 𝑘) = \binom{n}{k}p^{k}(1−p)^{n−k}

Taking the limit as 𝑛 → ∞, 𝑝 → 0, such that 𝑛𝑝 = 𝜆, we get:

𝑃(𝑋 = 𝑘) = \lim_{n→∞} \binom{n}{k}p^{k}(1−p)^{n−k} = \frac{e^{−𝜆}𝜆^{k}}{k!}

The Poisson distribution can be shown to arise from the Binomial distribution under certain conditions. Here, as the number of trials increases indefinitely and the probability of success decreases, we maintain a constant average number of successes (𝜆). When these conditions are satisfied, the complex Binomial formula simplifies and converges to the Poisson formula, allowing us to calculate probabilities of events in situations where events are rare.

Consider a large factory running an assembly line with thousands of products. The chance of a defective item is very small, say 0.001 per item. If you check many items, the situation mimics the Poisson distribution as more items are checked (infinity approach), leading to a manageable way to predict how many defects you might see in a smaller sample.

1. Poisson's Equation in PDEs: The Poisson equation is a PDE of the form: ∇²𝜙 = 𝑓(𝑥,𝑦,𝑧) It is used in problems involving electrostatics, gravitational fields, and heat conduction, where the source term 𝑓 is often related to a Poisson-distributed phenomenon.

2. Telecommunication: Models the number of phone calls or messages received per unit time.

3. Quality Control: Determines the number of defects in manufactured products.

4. Traffic Flow: Models vehicle arrivals at a traffic intersection.

5. Radiation Physics: Describes the number of radioactive decays in a given time frame.

The Poisson distribution finds various practical applications across engineering and science. For instance, in electrostatics, the Poisson equation models electric potential in a region given certain charge distributions. In communication fields, it helps predict call traffic. It is also vital in quality control, assessing how often defects appear in products. In traffic engineering, it can model how cars arrive at a stop. Lastly, in radiation physics, it predicts decay events of radioactive materials over time.

Imagine a cellphone network that needs to anticipate the number of incoming calls during peak hours. By applying Poisson distribution principles, engineers can effectively allocate resources to avoid overload. Similarly, in quality control, if a factory knows it averages 1 defect per 10 products, they can efficiently set quality checks based on this rate.

Feature Poisson Binomial Normal Type Discrete Discrete Continuous Domain 𝑘 = 0,1,2,… 𝑘 = 0,1,2,…,𝑛 −∞ < 𝑥 < ∞ Mean = Yes (𝜆) No No Variance Derived from Binomial (limit case) Central Limit Theorem Symmetry Skewed unless 𝜆 is large Approx. symmetric Symmetric

This comparison highlights how the Poisson distribution is distinct from both the Binomial and Normal distributions. The Poisson distribution is discrete, focusing solely on counting occurrences (like the number of failures), while the Binomial distribution also categorizes events based on the number of trials. The Normal distribution, in contrast, deals with continuous data. Understanding these differences is crucial for selecting the correct statistical approach depending on the data involved.

Think of cookie jars. The Poisson distribution helps count the number of chocolate chips in randomly chosen cookies, while the Binomial distribution helps track how many times a cookie comes out with chocolate chips when randomly sampled from the jar. The Normal distribution would be like measuring the average size of cookies baked in hundreds—dealing with a continuous range of sizes rather than discrete counts.

--