Probability Distributions

Probability distributions fall into two fundamental categories: discrete and continuous. A discrete distribution assigns probability to specific, countable outcomes. For example, the number of heads in 10 coin flips can only be 0, 1, 2, ..., 10. The probability of any specific value is given by the probability mass function (PMF). Discrete distributions are used when outcomes are distinct and countable.

A continuous distribution assigns probability to intervals of values rather than specific points. The probability density function (PDF) describes the relative likelihood of values. Since there are infinitely many possible values in any interval, the probability of any exact value is zero. Continuous distributions are used for measurements like height, time, temperature, and weight.

Understanding the distinction between discrete and continuous is essential for choosing the right distribution and the right calculator for your problem. A probability distribution calculator handles both types, but the inputs and outputs differ fundamentally between them.

Key Differences

• PMF vs PDF: PMF gives probabilities directly; PDF gives density that must be integrated over intervals
• CDF for both: The cumulative distribution function works the same way for both types: F(x) = P(X \u2264 x)
• Sum vs integral: Total probability sums to 1 for discrete (sum of PMF) and integrates to 1 for continuous (area under PDF)
• Histogram vs curve: Discrete distributions are visualized with bar charts; continuous with smooth curves

The binomial distribution models the number of successes in n independent trials, each with success probability p. It is the foundation of statistical hypothesis testing, quality control, and survey analysis. Parameters: n (number of trials), p (probability of success). Mean = np, Variance = np(1\u2212p), Support: k = 0, 1, ..., n.

Use the binomial distribution when you have a fixed number of independent yes/no trials with constant probability. For large n (n \u2265 30), the binomial distribution approximates the normal distribution with mean np and variance np(1\u2212p). For rare events (small p, large n), the Poisson approximation with \u03BB = np is more accurate.

Common applications: A/B testing results, defect rate analysis, survey response rates, clinical trial outcomes, sports win/loss records.

The Poisson distribution models the number of events occurring in a fixed interval when events happen independently at a constant average rate. It is uniquely characterized by the equality of its mean and variance, both equal to \u03BB. Parameters: \u03BB (rate parameter). Mean = \u03BB, Variance = \u03BB, Support: k = 0, 1, 2, ...

Use Poisson for count data where the variance approximates the mean. Common in epidemiology, queueing theory, insurance, and telecommunications. For large \u03BB (\u03BB \u2265 30), Poisson approximates the normal distribution with mean \u03BB and variance \u03BB.

Common applications: Call center arrivals per hour, website visits per day, insurance claims per month, particles detected per second.

The normal (Gaussian) distribution is the most important continuous distribution in statistics. Its symmetric bell-shaped curve appears throughout nature and is the foundation of the central limit theorem. Parameters: \u03bc (mean), \u03c3 (standard deviation). Mean = \u03bc, Variance = \u03c3\u00b2, Support: (\u2212\u221e, \u221e).

Use the normal distribution for continuous measurements that cluster around a central value. The 68-95-99.7 rule describes how data falls within 1, 2, and 3 standard deviations of the mean. The standard normal distribution (Z-distribution) with \u03bc = 0 and \u03c3 = 1 is used for z-score calculations.

Common applications: Height and weight distributions, IQ scores, measurement errors, test scores, natural phenomena.

The geometric distribution models the number of trials needed to achieve the first success in independent Bernoulli trials. It is the only discrete distribution with the memoryless property: past failures do not affect future waiting times. Parameters: p (probability of success per trial). Mean = 1/p, Variance = (1\u2212p)/p\u00b2, Support: k = 1, 2, 3, ...

The negative binomial distribution generalizes the geometric to model waiting time for the r-th success, with PMF P(X=k) = C(k\u22121, r\u22121) \u00d7 p^r \u00d7 (1\u2212p)^k\u2212r. Use geometric when you are waiting for a single success to occur in repeated trials.

Common applications: Free throw attempts until first make, items inspected until first defect, customer arrivals until first purchase.

The uniform distribution models outcomes where every value in the support is equally likely. It exists in both discrete and continuous forms and serves as the foundation for random number generation and simulation.

Discrete Uniform: P(X = x) = 1/(b\u2212a+1) for x = a, a+1, ..., b. Mean = (a+b)/2. Applications include fair dice rolls, random card draws, and lotteries with equal prize probabilities.

Continuous Uniform: f(x) = 1/(b\u2212a) for a \u2264 x \u2264 b. Mean = (a+b)/2, Variance = (b\u2212a)\u00b2/12. Applications include random number generation, random arrival times, and cases where all values in a range are equally plausible.

Both forms are essential for simulation studies, Monte Carlo methods, and as prior distributions in Bayesian statistics.

Selecting the correct probability distribution for your data is critical for accurate analysis. Start by asking whether your data is discrete or continuous. For discrete data, consider the underlying process: fixed number of trials (binomial), rare events over time (Poisson), or waiting time (geometric/negative binomial). For continuous data, consider symmetry (normal) or equal likelihood across a range (uniform).

The central limit theorem makes the normal distribution applicable in many situations where the sample size is large enough, regardless of the underlying distribution. However, for small samples or specialized problems, using the correct distribution is essential. Use our binomial probability calculator for yes/no trial counts, Poisson distribution calculator for rare event counts, and geometric probability calculator for waiting time problems.

Remember that distributions are models of reality, not reality itself. Always check your assumptions: independence of trials, constant probability, and appropriate support. When in doubt, compare multiple candidate distributions to your empirical data using goodness-of-fit tests or visualization.

Quick Selection Guide

• Yes/No outcomes, fixed trials \u2192 Binomial
• Rare events, constant rate \u2192 Poisson
• Continuous measurements, symmetric \u2192 Normal
• Waiting for first success \u2192 Geometric
• Waiting for r-th success \u2192 Negative Binomial
• Equally likely outcomes \u2192 Uniform