How to read it 01 Pick a family — the parameter fields and their defaults follow the tab. 02 Enter x or k and read the density/mass, the CDF and the tail card. 03 Set the inverse p. The quantile card returns the value whose CDF reaches p; for discrete families that is the smallest whole k. 04 Hand the number on. The quantile or the point value can be sent to the tests page or published as a card.
Worked examples
Family · input
Readouts
Poisson · λ = 4, k = 6
PMF 0.104196 , CDF 0.889326 , P(X ≥ 6) 0.21487 ; quantile at p = 0.95 → 8
Geometric · p = 0.3, k = 3
PMF 0.147 , CDF 0.657 , P(X ≥ 3) 0.49 ; mean 3.33333 , variance 7.77778
F · d1 = 5, d2 = 10, x = 2
PDF 0.162006 , CDF 0.835805 , right tail 0.164195 ; 95th percentile 3.32583
χ² · df = 10, x = 2
PDF 0.00766416 , CDF 0.00365985 , right tail 0.99634 ; 95th percentile 18.307
Binomial · n = 20, p = 0.5, k = 12
PMF 0.120134 , CDF 0.868412 , P(X ≥ 12) 0.251722 ; mean 10 , variance 5
The Poisson quantile moves in steps, not smoothly: CDF(7) is 0.948866 and CDF(8) is 0.978637 , so p = 0.95 lands on 8 .
Limits to keep in mind
Six significant digits per card. Values print short; the quantile rounds the same way.
A discrete quantile is a step, not an interpolation. The answer is the smallest integer whose CDF reaches p, which can sit well away from a p × support guess. What it does not do. It neither fits parameters to a sample nor draws random values: no estimation, no goodness-of-fit, no simulation. Inference about a sample is on the tests page; lifetime distribution fitting sits in Applied Statistics .
The geometric tab counts trials The geometric family uses the trial-index convention: X is the number of the trial that produces the first success, so X ≥ 1, PMF(0) reads 0 , and at p = 0.3 the third trial carries P(X = 3) = 0.147 . Under the other common reading — failures before the first success — the same event is k = 2 with probability 0.147 , while k = 2 under the trial-index reading (success on trial 2) is 0.21 ; the median flips from 2 trials to 1 failure. Both conventions are standard; every number on this tab commits to the trial-index one. The mean 3.33333 and variance 7.77778 are the cross-check when a source does not say which convention it used.
Where it helps Sizing for a burst With λ = 4 arrivals per interval, P(X ≥ 6) = 0.21487 : roughly one interval in five carries six or more. The 95th percentile 8 is the level to provision for, and changing λ shows how fast the tail moves before capacity is fixed.
Pulling critical values for tables The inverse field works as a small critical-value table: the 95th percentile is 3.32583 for F(5, 10) and 18.307 for χ²(10). The tests page walks the other direction, turning a statistic into a p-value.
Privacy Densities, tails and quantiles are computed in your browser; parameters you type never leave the page.
References
NIST/SEMATECH, e-Handbook of Statistical Methods , §1.3.6 Probability Distributions, itl.nist.gov (访问日期:2026-10-01)— distribution families, tails and critical regions.
Wikipedia, Poisson distribution , en.wikipedia.org (访问日期:2026-10-01)— arrivals and the λ mean/variance.
Wikipedia, Geometric distribution , en.wikipedia.org (访问日期:2026-10-01)— the two support conventions.
Wikipedia, F-distribution , en.wikipedia.org (访问日期:2026-10-01)— the d1, d2 ratio family behind ANOVA.
Wikipedia, Chi-squared distribution , en.wikipedia.org (访问日期:2026-10-01)— the df family behind goodness-of-fit.
Calculators in this hub
Hand-picked tools, one click away. The mini versions compute live and carry your values into the full calculator.
Sources & review
Reviewed by CalcX Editorial Team
Updated 2026-10-01