Skip to main content

Function reference overview

All 543 functions registered by duckfn_statrs share the sr_ prefix. Every function in the sub-pages below includes at least one executable SQL example that runs directly in your browser.

Search all registered functions at runtime:

SELECT function_name, function_type, description
FROM duckdb_functions()
WHERE function_name LIKE 'sr_%'
ORDER BY function_name;

How this section is organised​

The sidebar mirrors the function families, with distributions split by kind:

  • Summary statistics — descriptive aggregates and the empirical distribution.
  • Distributions
    • Continuous — normal / log-normal, the gamma family, beta, location-scale, extreme value and heavy tail, Student's t and Fisher-Snedecor, uniform / triangular / Dirac.
    • Discrete — Bernoulli and binomial trials, Poisson and hypergeometric, categorical and discrete uniform.
    • Multivariate — multivariate normal, Dirichlet and multinomial.
  • Special functions — error / gamma / beta families, factorials, harmonic numbers, logistic/logit, polynomial and kernel helpers (kernels are selected by name, e.g. sr_kernel_eval('gaussian', 0.0)) — plus the Constants: the mathematical constants and statrs' own floating-point precision thresholds with sr_almost_eq.
  • Random sampling — continuous, and discrete / multivariate / empirical samplers.
  • Density and signals, Hypothesis tests.

Each page gives every function its signature, parameter types and at least one executable SQL example that runs in your browser.

NULL semantics (applies everywhere)​

  • NULL input rows never enter an aggregate; scalars short-circuit to NULL.
  • Results that statrs cannot define (empty group, insufficient samples, out-of-range parameter) become SQL NULL — never NAN.
  • A parameter that is present but invalid (e.g. std_dev <= 0) raises a query error.

Common patterns​

All continuous distributions provide 5 functions per distribution:

sr_<name>_pdf(x, params...) probability density
sr_<name>_ln_pdf(x, params...) log-density
sr_<name>_cdf(x, params...) cumulative distribution P(X <= x)
sr_<name>_sf(x, params...) survival function P(X > x)
sr_<name>_quantile(p, params...) inverse CDF

All discrete distributions substitute pmf/ln_pmf for pdf/ln_pdf.

On top of those five, each distribution also exposes the quantities statrs can define — mean, variance, standard deviation, entropy, skewness, median, mode and the support bounds:

sr_<name>_mean(params...) distribution mean
sr_<name>_variance(params...) variance
sr_<name>_std_dev(params...) standard deviation
sr_<name>_entropy(params...) differential / Shannon entropy
sr_<name>_skewness(params...) skewness
sr_<name>_median(params...) median
sr_<name>_mode(params...) mode
sr_<name>_min(params...) lower support bound
sr_<name>_max(params...) upper support bound

Where statrs has no closed form (or none exists), the function is still registered and returns SQL NULL — for example the mean and variance of a Cauchy distribution.