Continuous versus discrete
Normal, gamma, beta and exponential models use a density function (PDF). Probabilities are areas under a curve. Binomial, Poisson and geometric models use a probability mass function (PMF) for individual integer outcomes.
STATISTICS LAB / INTERACTIVE EXPLORER
Understand distributions visually. Adjust parameters, measure probabilities, compare curves, and simulate outcomes — entirely in your browser.
Probability density and selected region
The filled area represents the selected probability. The vertical axis is probability density, not probability, for continuous distributions.
Select a region or solve for a percentile.
Continuous intervals include their endpoints; endpoints have zero probability individually.
Presets use equal-tail quantiles; for discrete outcomes, actual coverage may exceed the selected target.
Generate data from the selected distribution using a deterministic seed.
Expected bin counts sum only over the displayed sample range. This does not imply a fitted distribution.
| Interval | Observed | Expected |
|---|
Normal, gamma, beta and exponential models use a density function (PDF). Probabilities are areas under a curve. Binomial, Poisson and geometric models use a probability mass function (PMF) for individual integer outcomes.
Use normal for approximately symmetric measurements; binomial for successes in fixed independent trials; Poisson for counts at a rate; exponential for waiting time; and Student’s t when modeling heavy-tailed sampling variation.
The cumulative distribution function (CDF) returns P(X ≤ x). A percentile finds the smallest value with at least the requested cumulative probability. Discrete percentiles step between integers, unlike smooth continuous quantiles.
This is an educational probability calculator, not a hypothesis-testing or financial risk certification tool. Extremely small tail probabilities can be limited by floating-point precision. Sample generation is deterministic for exploration, not cryptographically secure.
Explore supported probability distributions with PDF/PMF, CDF, interval and tail queries, seeded samples, and exportable charts or tables.
Choose a normal distribution with a fixed mean and spread, compare a central interval with a tail query, then generate a seeded sample and export the chart and summary.
These are bounded double-precision calculations, not fitted models or statistical guarantees. Discrete outside queries exclude endpoints. Equal-tail intervals may differ from target mass, and histogram bins restricted to the observed sample range can omit out-of-range sample values. Parameter changes invalidate old simulations. Check the selected distribution and assumptions before interpreting results.
No. The explorer calculates a selected theoretical distribution; it does not fit or validate a model against your data.
No. A seed makes the generated sequence repeatable, not representative or statistically guaranteed.