dzibinom: Zero-inflated binomial probability density
In joepowers16/rethinking: Statistical Rethinking book package

Description Usage Arguments Details Author(s)

Functions for computing density and producing random samples from a zero-inflated binomial probability distribution.

1 2	dzibinom( x , p_zero , size , prob , log=FALSE ) rzibinom( n , p_zero , size , prob )

`x`	Integer values to compute densities or probabilies of
`p_zero`	Probability of a zero from bernoulli process
`size`	Number of binomial trials
`prob`	Probability of success in binomial trial
`log`	If `TRUE`, returns log-probability instead of probability

These functions provide density and random number calculations for zero-inflated binomial observations. This distribution is defined as a finite mixture of zeros and binomial values, where p_zero determines the weight of the pure zeros. As such, the probability of a zero is p_zero + (1-p_zero)(1-prob)^size. And the probability of a non-zero value x is (1-p_zero) choose(size,x) prob^x (1-prob)^(size-x).

dzibinom does its calculations in a way meant to reduce numerical issues with probabilities close to 0 and 1. As a result, it's not very fast.

Richard McElreath

joepowers16/rethinking documentation built on June 2, 2019, 6:52 p.m.