flexmix (version 0.7-1)

flexmix: Flexible Mixture Modeling

Description

FlexMix implements a general framework for finite mixtures of regression models. Parameter estimation is performed using the EM algorithm: the E-step is implemented by flexmix, while the user can specify the M-step.

Usage

flexmix(formula, data = list(), k = NULL, cluster = NULL,
        model=NULL, control = NULL)

Arguments

formula

A symbolic description of the model to be fit. The details of model specification are given below.

data

An optional data frame containing the variables in the model.

k

Number of clusters (not needed if cluster is specified).

cluster

Factor or integer vector with the initial cluster assignments of observations at the start of the EM algorithm (default is random assignment into k clusters).

model

Object of FLXmodel of list of FLXmodel objects. Default is the object returned by calling FLXglm().

control

Object of class FLXcontrol or a named list.

Value

• Returns an object of class flexmix.

Details

FlexMix models are described by objects of class FLXmodel, which in turn are created by driver functions like FLXglm or FLXmclust. Multivariate responses with independent components can be specified using a list of FLXmodel objects.

See Also

flexmix-class, FLXglm, FLXmclust

Examples

## generate some artificial data with a Gaussian response y and a
## Poisson response yp
x <- runif(200, 0, 10)
mydf <- data.frame(x=x,
                 y=c(5*x[1:100], 40-(x[101:200]-5)^2)+3*rnorm(200),
                 yp=rpois(200, exp(c(2-0.2*x[1:100], 1+0.1*x[101:200]))),
                 class=rep(1:2, c(100,100)))

plot(y~x, data=mydf, col=class)

## mixture of two linear regression models. Note that control parameters
## can be specified as named list and abbreviated if unique.
ex.qpol <- flexmix(y~x+I(x^2), data=mydf, k=2,
                   control=list(verb=5, iter=100))

ex.qpol
ex.qpol@components

## now use both responses simultanously
## first we have a look at the data
opar <- par("mfrow")
par(mfrow=1:2)
plot(y~x, data=mydf, col=class)
plot(yp~x, data=mydf, col=class)
par(mfrow=opar)

## now we fit a model with one Gaussian response and one Poisson
## response. Note that the formulas inside the call to FLXglm are
## relative to the overall model formula.
ex.qpol1 <- flexmix(y~x, data=mydf, k=2,
                    model=list(FLXglm(y~.+I(x^2)), 
                               FLXglm(yp~., family="poisson")))  

ex.qpol1
table(ex.qpol1@cluster, mydf$class)

## for Gaussian responses we get coefficients and standard deviation
parameters(ex.qpol1, component=1, model=1)

## for Poisson response we get only coefficients
parameters(ex.qpol1, component=1, model=2)