mixBinom: EM algorithm for estimating binomial mixture model

In davismcc/cardelino: Clone Identification from Single Cell Data

EM algorithm for estimating binomial mixture model

Description

EM algorithm for estimating binomial mixture model

Usage

mixBinom(
  k,
  n,
  n_components = 2,
  p_init = NULL,
  learn_p = TRUE,
  min_iter = 10,
  max_iter = 1000,
  logLik_threshold = 1e-05
)

Arguments

k	A vector of integers. number of success
n	A vector of integers. number of trials
n_components	A number. number of components
p_init	A vector of floats with length n_components, the initial value of p
learn_p	bool(1) or a vector of bool, whether learn each p
min_iter	integer(1). number of minimum iterations
max_iter	integer(1). number of maximum iterations
logLik_threshold	A float. The threshold of logLikelihood increase for detecting convergence

Value

a list containing p, a vector of floats between 0 and 1 giving the estimated success probability for each component, psi, estimated fraction of each component in the mixture, and prob, the matrix of fitted probabilities of each observation belonging to each component.

Examples

n1 <- array(sample(1:30, 50, replace = TRUE))
n2 <- array(sample(1:30, 200, replace = TRUE))
k1 <- apply(n1, 1, rbinom, n = 1, p = 0.5)
k2 <- apply(n2, 1, rbinom, n = 1, p = 0.01)
RV <- mixBinom(c(k1, k2), c(n1, n2))

davismcc/cardelino documentation built on Nov. 19, 2022, 2:44 a.m.