91Ó°ÊÓ

The hypergeometric distribution is fundamental when you're dealing with smaller populations and specific successes. Imagine you have a small group of 50 firms, where 15 are breaking some kind of regulations. When an inspector selects 10 of these firms randomly, the exact distribution for this scenario is hypergeometric. This relies on the precise number of choices and items you're drawing without replacement.

In simple terms, it helps answer the question: "What’s the probability of picking a certain number of violators from those 10 firms?" This kind of problem is frequent in quality control and other real-world settings where the sample is a notable fraction of the total population.

To find this probability mass function (pmf) in our case, we'd use the formula:

• \[ P(X = k) = \frac{{\binom{15}{k} \binom{35}{10-k}}}{\binom{50}{10}} \]

Here, \(k\) is the number of violating firms we find. By applying this formula for each \(k\) (from 0 to 10), you can compute each possible outcome's probability. This meticulous breakdown reveals just how likely any specific number of rule-breakers is in your sample.

The binomial distribution offers a handy approximation tool when working with large populations. Let's say instead of 50, we have 500 companies, 150 of which are known violators. Inspecting only a small sample, like 10, allows us to swap to a binomial model rather smoothly. This transition happens because of the reduced impact any single sample makes on such a large population.

For binomial distribution, each company visit is a trial. The primary question is whether this firm is a violator or not. The probability of selecting a violator in this expansive scenario boils down to the ratio \( p = \frac{150}{500} = 0.3 \).

The probability mass function (pmf) for the binomial distribution is simpler:

• \[ P(X = k) = \binom{10}{k}(0.3)^k(0.7)^{10-k} \]

This binomial view simplifies calculations notably, because every trial is independent, and the total sample size is small relative to the total population. This technique is common in statistics, especially when the total population size is overwhelming, but you still want to understand probabilities within a reasonably sized sample.

When dealing with probability distributions, understanding the expected value (mean) and variance is crucial for predicting outcomes.

For the hypergeometric distribution, where you have a specific number of firms and violators, the expected value \(E(X)\) tells us the average number of violators expected among the picked firms. In our example, this calculates to:

• \[ E(X) = \frac{10 \times 15}{50} = 3 \]

This indicates that, on average, 3 of the 10 visited firms will be violating regulations. The variance here shows the spread or how much the outcomes differ from this average:

• \[ V(X) = 10 \times \frac{15}{50} \times \frac{35}{50} \times \frac{50-10}{50-1} = 1.8367 \]

Switching to the binomial distribution approximation, the expected value \(E(X)\) remains the same:

• \[ E(X) = np = 10 \times 0.3 = 3 \]

This consistent average reassures us of its accuracy. The variance, however, slightly changes because of distribution differences

• \[ V(X) = np(1-p) = 10 \times 0.3 \times 0.7 = 2.1 \]

Despite this variance variation, both distributions give valuable insights into differing probability scenarios. Understanding how both the hypergeometric and binomial distributions affect expected value and variance helps in interpreting real-world problems effectively.