fmbeg_em: EM algorithm function for the mixed bivariate distribution...

In camponsah/BivMixDist: Fit bivariate continous and discrete distributions as well as provide EM algorithm for discrete Pareto

Description

This function computes the parameter estimates of FMBEG distribution using the EM algorithm.

Usage

fmbeg_em(
  data,
  beta = NULL,
  p = NULL,
  q = NULL,
  pi = NULL,
  m = 2,
  l = 1,
  maxiter = 1000,
  tol = 1e-08,
  verb = FALSE
)

Arguments

data	dataframe of bivariate random vector (X,N) BMEG distribution.
beta	numeric parameter vectors, which must be numeric greater than 0. Default value is NULL.
p	vector of numeric parameters between 0 and 1. Default value is NULL.
q	vector of gamma membership probabilities between 0 and 1, and sum equal to 1. Default value is NULL.
pi	vector of geometric membership probabilities between 0 and 1, and sum equal to 1. Default value is NULL.
m	number of gamma components. Default value is 2.
l	number of geometric components. Default value is 1.
maxiter	maximum number of iterations. Default value is 1000.
tol	tolerance value. Default value is 1e-08
verb	If TRUE, estimates are printed during each iteration.

Details

Takes initial guess for the parameters and the algorithm will estimate the MLE.

Value

list containing data frame parameter estimates, log-likelihood value and data frame of iteration

References

Amponsah, C. K. and Kozubowski, T.J., and Panorska, A.K. (2020). A Mixed bivariate distribution with mixture exponential and geometric marginals. Inprint.

Examples

data <- rfmbeg(10, beta = c(1,2,10), p=c(0.5,0.8), q=c(0.3,0.2,0.5), pi=c(0.6,0.4))
fit <- fmbeg_em(data=data, m=3,l=3)
fit$par

camponsah/BivMixDist documentation built on Nov. 15, 2021, 3:11 a.m.