Probability distributions

The sktime.distributions module contains probability distributions which combine a pandas.DataFrame-like API with a scikit-base compatible object interface.

All distributions in skpro can be listed using the skpro.registry.all_objects utility, using object_types="distribution", optionally filtered by tags. Valid tags can be listed using sktime.registry.all_tags.

Base

BaseDistribution([index, columns])

Base probability distribution.

Parametric distributions

Continuous support

Fisk([alpha, beta, index, columns])	Fisk distribution, aka log-logistic distribution.
Laplace(mu, scale[, index, columns])	Laplace distribution.
Logistic(mu, scale[, index, columns])	Logistic distribution.
Normal(mu, sigma[, index, columns])	Normal distribution (skpro native).
TDistribution(mu, sigma[, df, index, columns])	Student's t-distribution (skpro native).
Weibull(scale, k[, index, columns])	Weibull distribution.

Integer support

Poisson(mu[, index, columns])

Poisson distribution.

Non-parametric and empirical distributions

Empirical(spl[, weights, time_indep, index, ...])	Empirical distribution (skpro native).
QPD_Empirical(quantiles[, time_indep, ...])	Empirical quantile parametrized distribution.
QPD_U(alpha, qv_low, qv_median, qv_high[, ...])	Johnson Quantile-Parameterized Distributions with unbounded mode.
QPD_S(alpha, qv_low, qv_median, qv_high[, ...])	Johnson Quantile-Parameterized Distributions with semi-bounded mode.
QPD_B(alpha, qv_low, qv_median, qv_high, ...)	Johnson Quantile-Parameterized Distributions with bounded mode.

Composite distributions

Mixture(distributions[, weights, index, columns])

Mixture of distributions.