R/tTest.R

In nrauscher/corpus: Toolkit for Corpus Analysis

#' perform t-test
#' 
#' S4 method for context object to perform t-test
#' @param .Object a context or comp object
#' @rdname tTest
#' @name tTest
setGeneric("tTest", function(.Object){standardGeneric("tTest")})

#' @rdname tTest
setMethod("tTest", "context", function(.Object){
  pRandom <- (.Object@stat[["count_partition"]] / .Object@partitionSize) * ( .Object@count / .Object@partitionSize)
  pSample <- .Object@stat[["count_window"]] / .Object@partitionSize
  tValues <- (pSample - pRandom) / sqrt( pSample / .Object@partitionSize )
  .Object@stat[, "t" := tValues]
  .Object <- sort(.Object, by="t")
  .Object@stat[, "rank_t" := c(1:nrow(.Object@stat))]
  .Object@method <- c(.Object@method, "t")
  return(.Object)
})