See Details for a description of the individual functions.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20  ## S3 method for class 'mixmod'
logLik(object, ...)
## S3 method for class 'mixmod'
qval(x, map=FALSE, ...)
## S3 method for class 'mixmod'
qfun(x, map=FALSE, ...)
## S3 method for class 'mdmixmod'
qfun(x, map=FALSE, ...)
## S3 method for class 'mixmod'
aic(x, ...)
## S3 method for class 'mixmod'
bic(x, ...)
## S3 method for class 'mixmod'
entropy(x, map=FALSE, ...)
## S3 method for class 'mixmod'
iclbic(x, map=FALSE, ...)
## S3 method for class 'mdmixmod'
siclbic(x, map=FALSE, ...)

x, object 
an object of class 
map 
logical; if 
... 
currently unused. 
logLik
calculates L(thetaX), the loglikelihood of the estimated parameters theta with respect to the observed data X, while qval
calculates the “Qvalue”, the expectation with respect to the hidden data of the loglikelihood with respect to the complete data: Q(theta) = E[L(thetaX,Y)] for mixmod
and Q(theta) = E[L(thetaX,Y,Y0)] for mdmixmod
. qfun
returns the hidden and observed portions of the Qvalue separately, as elements of a vector.
aic
, bic
, entropy
, iclbic
, and siclbic
calculate various information criteria for model selection with mixture models of class mixmod
and mdmixmod
. These criteria are Akaike's information criterion (AIC, Akaike, 1974), the Bayes information criterion (BIC, Schwarz, 1978), the classification entropy (Biernacki et al., 2000), the integrated complete likelihood BIC (ICLBIC, Biernacki et al., 2000), and the simplified ICLBIC (SICLBIC) for objects of class mdmixmod
, respectively. They are defined as follows:
AIC  =  2 L(thetaX)  2 Theta 
BIC  =  2 L(thetaX)  Theta log(N) 
entropy  =  2 L(thetaX)  2 Q(theta) 
ICLBIC  =  2 Q(theta)  Theta log(N) 
SICLBIC  =  2 E[L(thetaX,Y0)]  Theta log(N) (mdmixmod only)

where Theta is the size of the parameter space and N is the size of the data. Generally, the model which provides the highest value of any information criterion should be selected. Current testing indicates that ICLBIC is preferred for mixmod
and BIC for mdmixmod
.
A numeric vector for qfun
, a numeric scalar for the other functions.
Some authors define AIC, BIC, and ICLBIC as the negative of the quantities given in Details.
Daniel Dvorkin
Akaike, H. (1974) A new look at the statistical model identification, IEEE Transactions on Automatic Control 19(6), 716–723.
Biernacki, C. and Celeux, G. and Govaert, G. (2000) Assessing a mixture model for clustering with the integrated completed likelihood, IEEE Transactions on Pattern Analysis and Machine Intelligence 22(7), 719–725.
McLachlan, G.J. and Thriyambakam, K. (2008) The EM Algorithm and Extensions, John Wiley & Sons.
Schwarz, G. (1978) Estimating the dimension of a model, The Annals of Statistics 6(2), 461–464.
mixmod
and mdmixmod
for details of the hidden variable structure.
1 2 3 4 5 6 7 8 
Questions? Problems? Suggestions? Tweet to @rdrrHQ or email at ian@mutexlabs.com.
All documentation is copyright its authors; we didn't write any of that.