R/getStats.R

In arorarshi/survClust: Identification Of Clinically Relevant Genomic Subtypes Using Outcome Weighted Learning

#' Compute fit statistics after cross validation via \code{cv_survclust}
#'
#' @param cv.fit output from \code{cv_survclust} object 
#' @param kk number of \code{k} clusters on which \code{cv_survclust} was run, default is \code{8}
#' @param cvr round of cross-validation on which \code{cv_survclust}  was run, default is \code{50} 
#' @details \code{getStats} calculates Logrank statistic and standardized pooled within sum of squares (SPWSS) across 
#' cross-validated labels. Visualize it via \code{plotStats}
#' @return A list of the following
#' \itemize{
#'  \item{lr} {log rank statistic}
#'  \item{spwss} {standardized pooled within sum of squares}
#'  \item{bad.sol} {number of solutions for each \code{kk} that have cluster class \code{<5} samples}
#'  }
#' 
#' @examples 
#' library(survClust)
#' ss_stats <- getStats(uvm_survClust_cv.fit, kk=7, cvr=10)
#' 
#' @author Arshi Arora
#' @export
getStats <- function(cv.fit,kk=8, cvr=50){
    kk <- kk-1
    cvr <- cvr
    lr <- matrix(NA, nrow=cvr, ncol=kk)
    spwss <- matrix(NA, nrow=cvr, ncol=kk)
    bad.sol <- rep(NA, length=kk)
    
    for(i in seq_len(kk)){
        lr[,i] <- t(unlist(lapply(cv.fit[[i]], function(x) x$cv.logrank)))
        spwss[,i] <- t(unlist(lapply(cv.fit[[i]], function(x) x$cv.spwss)))
        min.labels <- unlist(lapply(cv.fit[[i]], function(x) min(table(x$cv.labels)) [1]))
        idx <- which(min.labels <=5)
        #if (length(idx) !=0){llk = llk[idx]
        #    print(paste0("k= ", i, " has few samples in cluster solution for CV round - ",idx))}
        bad.sol[i] <- length(idx)
        
    }
    return(list(lr=lr, spwss = spwss, bad.sol = bad.sol))
}