Saved

Press the star next to a tool's name to save it.

All my tools

Probability Distribution Calculator

Use this probability distribution calculator for the normal, lognormal, exponential, Weibull, Rayleigh, uniform, beta, Poisson, geometric, negative binomial, hypergeometric and binomial.

Restored your last values ·

Choose one of twelve distributions, enter its parameters, and choose what to find: a cumulative probability, the probability or density at a value, an upper tail, a probability between two values, or the value for a given probability. The result comes with the mean and variance. For the binomial, the normal approximation is shown too.

Everything is worked out in your browser. Nothing you type is sent anywhere.

–

–

Choose a distribution and what to find, fill in the boxes, then press Calculate.

A probability distribution calculator finds probabilities, densities and quantiles for twelve common distributions: normal, lognormal, exponential, Weibull, Rayleigh, uniform, beta, Poisson, geometric, negative binomial, hypergeometric and binomial. Choose a distribution, enter its parameters and the value, and you get the result with the mean and variance. For the binomial it also gives the normal approximation with a continuity correction.

How to use the probability distribution calculator

  1. Choose the distribution and enter its parameters. The labels change to match the distribution you picked.
  2. Choose what to find: P(X ≤ x), P(X = x), an upper tail, the probability between two values, or the value that matches a probability.
  3. Enter x, or two values, or the probability, and press Calculate.

The distributions and their parameters

The continuous distributions are: normal, with mean μ and standard deviation σ; lognormal, where μ and σ describe the natural logarithm of the variable, not the variable itself; exponential, with rate λ; Weibull, with scale λ and shape k; Rayleigh, with scale σ; uniform, between a and b; and beta, with shapes α and β on the interval from 0 to 1. The discrete ones are: Poisson, with rate λ; geometric, with the success probability p and x counting the trials up to the first success; negative binomial, with r successes needed and success probability p; hypergeometric, with population N, K successes in it and n draws; and binomial, with n trials and probability p. The density, cumulative and quantile formulas of the continuous distributions are the standard textbook ones, which we give as our own working.

The negative binomial

The Wikipedia article on the negative binomial distribution warns that definitions differ in whether the support starts at k = 0 or at k = r, and in whether p is the probability of success or failure.[3] The calculator uses the primary definition on the page: the number of failures before the r-th success, with success probability p.[3] The page gives the cumulative probability as the regularized incomplete beta function I_p(r, k + 1), the mean as r(1 − p) ÷ p and the variance as r(1 − p) ÷ p².[3] With r = 5 and p = 0.4 the mean is 7.5 failures. The page's candy-selling example with r = 5 and p = 0.4 gives a probability of about 0.10033 for a total of 10 houses visited, which is 5 failures in our counting; enter x = 5 to see it.[3]

The hypergeometric

The hypergeometric distribution describes draws without replacement. The page writes it as Hypergeometric(N, K, n) and gives the mean as nK ÷ N.[4] Its example is N = 50, K = 5 and n = 10, with P(X = 4) of about 0.003964583 and P(X = 5) of about 0.0001189375.[4] Enter those values with the probability option to reproduce them. The calculator accepts only whole numbers for N, K and n, with K no larger than N and n no larger than N.

The binomial and its normal approximation

The article on the binomial distribution gives the mean as np and the variance as np(1 − p).[2] It says that for large n the distribution is approximated by the normal distribution with that mean and variance, improved by a continuity correction, and that the approximation generally improves as n increases and is better when p is not near 0 or 1.[2] A common rule is that np and n(1 − p) must both be at least 5, though the page says the number varies from source to source.[2] The calculator shows a caution when either is below 5. The page's example approximates P(X ≤ 8) by the normal probability up to 8.5.[2]

The continuity correction

The article on the continuity correction explains that for whole numbers P(X ≤ x) equals P(X < x + 1), and that if np and np(1 − p) are large, sometimes taken as both at least 5, then P(X ≤ x) is approximately the normal probability P(Y ≤ x + 1/2).[1] The page adds that the same idea applies to a Poisson variable, using a normal with mean and variance λ.[1] The page gives only this rule. For an upper tail, P(X ≥ k) is approximately P(Y ≥ k − 1/2), and for a single value, P(X = k) is approximately P(k − 1/2 ≤ Y ≤ k + 1/2). These two follow from the first, and they are our own working. For n = 20 and p = 0.5, the exact P(X ≤ 8) is 0.2517, the corrected normal value is 0.2512 and the uncorrected value is 0.1855, our own calculations that show how much the correction helps.

Quantiles

For the inverse, enter a probability and the calculator finds the value x with P(X ≤ x) equal to that probability for continuous distributions. For the normal, lognormal, exponential, Weibull, Rayleigh and uniform distributions it uses the closed form, and for the beta it searches numerically. For a discrete distribution, there is usually no x with exactly that probability, so the calculator gives the smallest whole number whose cumulative probability is at least the probability that you entered, which is our own convention. For the normal with mean 0 and standard deviation 1, a probability of 0.975 gives 1.95996.

Choosing a distribution

Use the normal for symmetric measurements, the lognormal for positive skewed quantities, the exponential for waiting times between events, the Weibull for lifetimes, the beta for proportions, the Poisson for counts of rare events, the geometric and negative binomial for counting trials until successes, the hypergeometric for sampling without replacement and the binomial for a fixed number of independent trials. Check that your problem matches the assumptions before you rely on a probability.

Common mistakes

  • Using the mean instead of the rate. The exponential uses λ, which is the inverse of the mean.
  • Mixing the counting convention. The geometric counts trials, the negative binomial counts failures here.
  • Reading a density as a probability. A continuous density can exceed 1.

What next

Use this often? Add Probability to your Home screen.

Frequently asked questions

What is a probability distribution calculator?

A tool that finds probabilities, densities and quantiles for a chosen distribution from its parameters, and shows its mean and variance.

How do I find the inverse normal?

Choose the normal distribution, choose the value for a probability, and enter the mean, the standard deviation and the probability. For 0.975 on the standard normal, the value is 1.95996.

What is the continuity correction?

An adjustment of 0.5 when a discrete distribution is approximated by the normal. P(X ≤ x) is approximated by the normal probability up to x + 0.5.

When is the normal approximation to the binomial good?

When np and n(1 − p) are both at least 5, though the threshold varies by source. It improves as n grows.

What does the negative binomial count in this calculator?

The number of failures before the r-th success, with success probability p.

What went wrong?

We only receive the reason you pick, never what you typed into the tool.

Thanks for your feedback.

More Math Tools