mixnorm {MixSemiRob}

Parameter Estimation for Uni- or Multivariate Normal Mixture Models

Description

‘mixnorm’ is used to estimate parameters of a normal mixture model with equal variance. The function supports both one-dimensional and multi-dimensional data.

Usage

mixnorm(x, C = 2, sigma.known = NULL, ini = NULL, tol = 1e-05)

Arguments

x

an n by p matrix of observations where n is the number of observations and s is the dimension of data.

C

number of mixture components. Default is 2.

sigma.known

a vector or matrix of component standard deviations. Default is NULL, which means the standard deviations are unknown.

ini

initial values for the parameters. Default is NULL, which randomly sets the initial values using the given observations. If specified, it can be a list with the form of list(mu, pi, sigma), where mu is a vector of C component means, pi is a vector of C mixing proportions, and sigma is a vector of C component standard deviations (this element is only needed when sigma.known is not given).

tol

stopping criteria for the algorithm. Default is 1e-05.

Value

A list containing the following elements:

mu

estimated component means.

sigma

estimated component standard deviations. Only returned when sigma.known is not specified.

pi

estimated mixing proportions.

p

matrix containing estimated classification probabilities where the (i, j)th element is the probability of the jth observation belonging to the ith component.

lik

final likelihood.

See Also

complh, distlat

Examples

# See examples for the `complh' function.