ZINegativeBinomial {distributions3}R Documentation

Create a zero-inflated negative binomial distribution


Zero-inflated negative binomial distributions are frequently used to model counts with overdispersion and many zero observations.


ZINegativeBinomial(mu, theta, pi)



Location parameter of the negative binomial component of the distribution. Can be any positive number.


Overdispersion parameter of the negative binomial component of the distribution. Can be any positive number.


Zero-inflation probability, can be any value in ⁠[0, 1]⁠.


We recommend reading this documentation on https://alexpghayes.github.io/distributions3/, where the math will render with additional detail.

In the following, let X be a zero-inflated negative binomial random variable with parameters mu = \mu and theta = \theta.

Support: \{0, 1, 2, 3, ...\}

Mean: (1 - \pi) \cdot \mu

Variance: (1 - \pi) \cdot \mu \cdot (1 + (\pi + 1/\theta) \cdot \mu)

Probability mass function (p.m.f.):

P(X = k) = \pi \cdot I_{0}(k) + (1 - \pi) \cdot f(k; \mu, \theta)

where I_{0}(k) is the indicator function for zero and f(k; \mu, \theta) is the p.m.f. of the NegativeBinomial distribution.

Cumulative distribution function (c.d.f.):

P(X \le k) = \pi + (1 - \pi) \cdot F(k; \mu, \theta)

where F(k; \mu, \theta) is the c.d.f. of the NegativeBinomial distribution.

Moment generating function (m.g.f.):

Omitted for now.


A ZINegativeBinomial object.

See Also

Other discrete distributions: Bernoulli(), Binomial(), Categorical(), Geometric(), HurdleNegativeBinomial(), HurdlePoisson(), HyperGeometric(), Multinomial(), NegativeBinomial(), Poisson(), ZIPoisson(), ZTNegativeBinomial(), ZTPoisson()


## set up a zero-inflated negative binomial distribution
X <- ZINegativeBinomial(mu = 2.5, theta = 1, pi = 0.25)

## standard functions
pdf(X, 0:8)
cdf(X, 0:8)
quantile(X, seq(0, 1, by = 0.25))

## cdf() and quantile() are inverses for each other
quantile(X, cdf(X, 3))

## density visualization
plot(0:8, pdf(X, 0:8), type = "h", lwd = 2)

## corresponding sample with histogram of empirical frequencies
x <- random(X, 500)
hist(x, breaks = -1:max(x) + 0.5)

[Package distributions3 version 0.2.1 Index]