'kmlcov' Cluster longitudinal data using the likelihood as a metric of distance. The generalised linear model allow the user to introduce covariates with different level effects (2 levels).
|Authors@R:||c(person("Mamoun O", "Benghezal", role = c("aut", "cre"), email = "email@example.com"), person("Christophe", "Genolini", role = c("ctb"), email = "firstname.lastname@example.org"), )|
|Collate:||'functions4glmClust.R' 'GlmCluster.R' 'glmclust-internal.R' 'glmClust.R' 'kmlcov-package.R' 'kmlCov.R'|
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
glmClust Clustering longitudinal data kmlCov Clustering longitudinal data from different starting conditions which_best Seek the best partitions Converge-class Class '"Converge"' GlmCluster-class Class '"GlmCluster"' KmlCovList-class Class '"KmlCovList"' addIndic Create the new formula with the indicator covariates affect_rand Affect randomly the individuals to the clusters getNomCoef Get the name of the coefficients in the 'glm' object according to the current cluster log_lik Calculate the log-likelihood lowcyclo Measures of creatinine and time among cardiac transplant patients. majIndica Calculate an indicator vector observance Measures of obsevance. plot,GlmCluster-method Plot the main trajectories plot,KmlCovList-method Plot the main trajectories predict_clust Creates a character string expression to calculate the predicted values rwFormula Rewrite the formula with all the covariates seperateFormula Separate the covariates in a formula show,Converge-method Method for function 'show'
To cluster longitudinal data, 'kmlcov' implement an ECM type algorithm
which assign the trajectories to the cluster which maximise the
It is possible to introduce covariates via the generalised linear model with different level effects (2 levels) all spedified in one formula.
The package implements the
plot function to produce the diagrams
at the condition of not having more than 2 different effects (although
the program can deal with more than two effects) for e.g.
sex q.v. the help of
linkglmClust or kmlCov. To plot the main trajectories with more
than two effects we recommand to use
ggplot of the package
To cluster longitudinal data, 2 functions have to be remembered
glmClust and kmlCov, the first run the algorithm for
clustering one time and the second run the same algorithm multiple times
with different starting conditions. The method is greatly sensitive to
the initial conditions, we therefore recommand to use
although it takes much more time.