Estimates the longitudinal concordance correlation to access the longitudinal agreement profile. The estimation approach implemented is variance components approach based on polynomial mixed effects regression model, as proposed by Oliveira, Hinde and Zocchi (2018)
lcc
is a package under development based on estimation procedures for longitudinal concordance correlation (lcc), longitudinal Pearson correlation (lpc), and longitudinal accuracy (la) through fixed effects and variance components of polynomial mixed-effect regression model. The main features of the package are its ability to perform inference about the extent of agreement and use a numerical and graphical to summary the fitted values, sampled values, and confidence intervals. Morever, our approach accommodate balanced or unbalanced experimental design, allows to model heteroscedasticity among within-group errors using or not the time as covariate, and also allows for inclusion of covariates in the linear predictor to control systematic variations in the response variable. It was developed by Thiago de Paula Oliveira [cre, aut], Rafael de Andrade Moral [aut], John Hinde [aut], Silvio Sandoval Zocchi [aut,ctb], Clarice Garcia Borges Demétrio [aut,ctb].
It has been available on CRAN since 2018 (https://cran.r-project.org/package=lcc). Its last version was updated on 2019-02-13. CRAN has lcc's stable version, which is recommended for most users.
This github page has its version under development. New functions will be added as experimental work and, once it is done and running correctly, we will synchronize the repositories and add it to the CRAN.
We worked hard to release a new stable version allowing users to analyze data sets, where the objective is studied the extent of the agreement profile among methods considering time as covariable.
lcc
comprises a set of functions that allows users build and summaries the fitted model, estimates and bootstrap confidence intervals for lcc, lpc and la statistics, and build graphical summaries for them. Some functions are used internally by the package, and should not be used directly.
install.packages("lcc")
install.packages("devtools")devtools::install_github("Prof-ThiagoOliveira/lcc")
If you use Windows, first install Rtools. If you are facing problems with Rtools installation, try to do it by selecting Run as Admnistrator option with right mouse button. On a Mac, you will need Xcode (available on the App Store).
lcc
can also be installed by downloading the appropriate files directly at the CRAN web site and following the instructions given in the section 6.3 Installing Packages
of the R Installation and Administration manual.
Under construction!! =D
Fixed issue when the wrong name of variable is declared.
Fixed issue when three or more levels of methods is considered in the dataset.
Fixed issue in lccWrapper function get the correct rho:
if(is.na(rho[[2]])=TRUE)
if(length(rho)==1){
return(rho[[1]])
}else(if(is.na(rho[[2]])){
return(rho[[1]])
}else(return(rho[[n.delta]])))
Fixed issue when the "all.plots" argument is declared FALSE. Now the list of plot can be split into multiple pages and save them as pdf using ggsave function. The marrangeGrob function of package gridExtra was used to solve this problem.
Fixed issue in lccWrapper function get the correct rho:
if(length(rho)==1){
return(rho[[1]])
}else(if(is.na(rho[[2]])){
return(rho[[1]])
}else(return(rho[[n.delta]])))
if(length(rho)==1){
return(rho[[1]])
}else(if(sum(is.na(rho[[2]]))!=0){
return(rho[[1]])
}else(return(rho[[n.delta]])))
We include the parameter 'type' in the lccPlot() function. Now the user can choice among lcc, lpc, and la as plot output. This allowed we reduce the number of parameters in the control list. Now we have only the parameters 'ylab' rather than 'LCC_ylab', 'LPC_ylab', and 'LA_ylab'. In the same way, we replace the parameters 'LCC_scale_y_continuous', 'LPC_scale_y_continuous', and 'LA_scale_y_continuous' by 'scale_y_continuous'.
Fixed issue when y-axis labels is changed.
A new parameter called interaction was included in the lcc() function. This parameter allows to estimate or not the interaction effect among predictors variables in the fixed part of the linear predictor.