1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
|
\name{vmm_fit_em}
\alias{vmm_fit_em}
\docType{methods}
\title{Estimate von Mises Mixture parameters using Expectation Maximization.}
\description{
Estimates parameters for univariate von Mises mixture using Expectation Maximization algorithm.
}
\usage{
vmm_fit_em( x, p, epsilon = c( 0.000001, 0.000001, 0.000001 ),
debug = FALSE, implementation = "C" )
}
\arguments{
\item{x}{data vector}
\item{p}{
initialization vector of 3*\emph{n} parameters, where \emph{n} is
number of mixture components. Structure of p vector is
p = c( A1, A2, ..., A\emph{n}, mu1, mu2, ..., mu\emph{n}, k1, k2, ..., k\emph{n} ),
where A\emph{i} is the proportion of \emph{i}-th component,
mu\emph{i} is the center of \emph{i}-th component and
k\emph{i} is the concentration of \emph{i}-th component.
}
\item{epsilon}{
tolerance threshold for convergence. Structure of epsilon is
epsilon = c( epsilon_A, epsilon_mu, epsilon_k ), where
epsilon_A is threshold for component proportions,
epsilon_mu is threshold for component centers and
epsilon_k is threshold for component concentrations.
}
\item{debug}{
flag to turn the debug prints on/off.
}
\item{implementation}{
flag to switch between C (default) and R implementations.
}
}
\value{
Vector of mixture parameters, whose structure is the same as of input
parameter's p.
}
\references{
Banerjee et al.
Expectation Maximization for Clustering on Hyperspheres (2003),
manuscript, accessible on:
\url{https://web.archive.org/web/20130120061240/http://www.lans.ece.utexas.edu/~abanerjee/papers/05/banerjee05a.pdf}
}
\author{Andrius Merkys}
|
