R/desc_table.R

In seonjoo/slmisc: Miscelaneous functions to conduct LeeLab projects

#' Create descriptive table by group
#' Current version works in the html setting. We are working on extend it to the word document
#'
#' @param varlist One-dimensional predictor
#' @param data Dataset to be analyzed.
#' @param groupvar Group variable name. If not specified, descriptive stats for all subjects are reported.
#' @param total (default=TRUE). Valid only if groupvar is not NULL. total=FALSE if a user does not want to print stats for all subjects.
#' @param group.comp (default=TRUE) Add p-values of group comparison. F test and Chi-square test are used for continuous and categorical variables, respectively.
#' @return squareroot of matrix
#' @examples
#' data(mtcars)
#' mtcars = mtcars %>% mutate(cyl=factor(cyl),vs=factor(vs), am=factor(am))
#' varlist=c('mpg','disp','hp','drat','wt','gear','carb','vs')
#' desc_table(varlist, data=mtcars,groupvar=NULL)
#' desc_table(varlist, data=mtcars,groupvar="cyl")
#' desc_table(varlist, data=mtcars,groupvar="cyl",total=FALSE)
#' @author Seonjoo Lee, \email{sl3670@cumc.columbia.edu}
#' @references TBA
#' @keywords highdimensional mediation L1penalization
#' @import dplyr
#' @import qwraps2
#' @import knitr
#' @export


desc_table<-function(varlist, data,groupvar=NULL, total=TRUE,group.comp=TRUE){

  ## determine variable types
  summary1<-list()
  vartype=unlist(lapply(varlist, function(x)ifelse(is.factor(data[,x])|is.character(data[,x]), 'cat','cont' )))
  for (j in 1:length(varlist)){

    if (vartype[j]=='cont'){
      summary1[[varlist[j] ]]<-list()
      eval(parse(text=paste('summary1[["',varlist[j],'"]][["',varlist[j],'"]] = ~ qwraps2::mean_sd(',varlist[j],
                            ', na_rm = TRUE,show_n="never")',sep='')))
    }
    if (vartype[j]=='cat'){
      catlist=names(table(data[,varlist[j]]))
      if(sum(is.na(data[,varlist[j]]))>0){
        data[,varlist[j]]=as.factor(ifelse(is.na(data[,varlist[j]]), 'NA' ,data[,varlist[j]]))
        catlist=names(table(data[,varlist[j]]))
      }
      eval(parse(text=paste("summary1[['",varlist[j],"']] =list(",
                            paste0(
                              paste("'",varlist[j],':',catlist,"'=~perc_n2(",varlist[j]," =='",catlist,"',na_rm=TRUE,show_denom='never')",sep=''),
                              collapse=','),
                            ")",sep='')))
    }
  }

### group comparison
  ps<-list()
  if (is.null(groupvar)==FALSE & group.comp==TRUE){
    for (j in 1:length(varlist)){
      if (vartype[j]=='cont'){eval(parse(text=paste("ps[[",j,"]]<-round(summary(aov(",varlist[j],"~",groupvar,",data))[[1]][1,5],4)")))}
      if (vartype[j]=='cat'){
        catlist=names(table(data[,varlist[j]]))
        tmp<-rep(" ",length(catlist))
        eval(parse(text=paste("tmp[1]<-round(chisq.test(table(data$",varlist[j],',data$',groupvar,'))$p.value,4)')))
        eval(parse(text=paste("ps[[",j,"]]<-tmp")))
        }
    }
  }

  whole<-summary_table(data, summary1)
  group<-c()
  if(is.null(groupvar)==FALSE){
    eval(parse(text=paste('group<-summary_table(dplyr::group_by(data,',groupvar,'), summary1)',sep='')))
    if (total==FALSE){whole<-c()}
  }

  if (is.null(groupvar)==FALSE & group.comp==TRUE){
    tmp.vec=unlist(ps)
    tmp.vec[tmp.vec=='0']<-'p<0.0001'
    group=cbind(group, tmp.vec)
  }
  both=cbind(whole,group)
  kable(both)
  #  return(both)

}