eexp_mix: Exponential Mixture Estimation

In matheuscastro43/finiteMix: Finite Mixture Models

Exponential Mixture Estimation

Description

Estimates parameters of a univariate Exponential mixture with k-means clustering and EM-algorithm.

Usage

eexp_mix(data, g, lim.em = 100, criteria = "dif.psi", epsilon = 1e-05, 
         plot.it = TRUE, empirical = FALSE, col.estimated = "orange", 
         col.empirical = "navy", ...)

Arguments

data	vector containing the sample, or list obtained with rexp_mix.
g	number of components in the mixture.
lim.em	limit of EM Iterations.
criteria	the stop criteria to be used, could be "dif.psi" to calculate differences on parameters matrix or "dif.lh" to calculate differences on Likelihood function.
epsilon	minimal difference value to algorithm stops.
plot.it	logical, TRUE to plot the histogram with estimated distribution curve.
empirical	logical, TRUE to add the empirical curve ("Kernel Density Estimation") in the plot.
col.estimated	a colour to be used in the curve of estimated density.
col.empirical	a colour to be used in the curve of empirical density.
...	further arguments and graphical parameters passed to hist.

Author(s)

CASTRO, M. O.; MONTALVO, G. S. A.

Examples

## Generate a sample.
data = rexp_mix(n = 1000, pi = c(0.2, 0.8), rate = c(1, 15))

## And now, estimate the parameters, using the 'data' list.
eexp_mix(data, g = 2)

## Or using the sample vector.
## like "iexp_mix(data$sample, g = 2)"

## Using the diference in the log-likelihood as stop criteria.
eexp_mix(data, g = 2, criteria = "dif.lh")

## Not plotting the graphic.
eexp_mix(data, g = 2, plot.it = FALSE)

## Adding the empirical curve to the graphic and changing the color of the 
## curves.
eexp_mix(data, g = 2, empirical = TRUE, col.estimated = "pink",
         col.empirical = "red3")

matheuscastro43/finiteMix documentation built on March 30, 2022, 12:49 p.m.