categorical functional data analysis Package: catfda
Type: Package
Title: categorical functional data analysis
Version: 0.1.0
Author: Xiaoxia Champon, Chathura Jayalath
Maintainer: The package maintainer [email protected]
URL: http://github.com/XiaoxiaChampon/catfda
BugReports: http://github.com/XiaoxiaChampon/catfda/issues
Description: This package contains three functions:
(1) Clustering: catfdcluster uses kmeans or DBSCAN on the multivariate
functional principal component scores extracted from the multivariate
latent curves which induce the categorical functional data. The scores
can be calculated using Ramsay's method or Happ's method.
(2) Hypothesis testing: aRLRTcfd uses approximate restricted likelihood
ratio test (aRLRT) to formally assess the relative importance of a category
for the class membership. The test is conducted directly on the working
normalized response where the parameters are estimated with penalized quasi-likelihood.
(3) Monitor clustering transition: catfdl incorporate the longitudinal nature of densely
observed categorical valued functional data and monitors the clustering transitions
over time through a single set of time varying functional principal component scores.
License:
Encoding: UTF-8
LazyData: true
Imports:
fda,
refund,
mgcv,
funData,
MFPCA,
dbscan,
fossil,
NbClust,
ggplot2
To use the functions in the package:
-
library(devtools)
install_github("XiaoxiaChampon/catfda")
-
library("catfda")
catfdclust=catfdcluster(matrix(sample(c(0,1,2),100*250,replace=TRUE),nrow=100,ncol=250),seq(0,1,length=250),25,3,3,0.9,4,5,2,"happ","two")
-
sample code is in the R folder: cluster_sample.R