R/base_n.r

In baileymh/Shuffle: What the Package Does (One Line, Title Case)

find_base_n.version = "1.5";

######Base n--------------------------------------------------

#' Find Base n
#'
#' This function finds the base n based on smallest number of participants (>10) needed for significance.
#' @param data_set Takes in a data.frame
#' @param alpha Alpha criteria, defaults to .05
#' @param paired Boolean to run Paired-Sample t-Test, defaults to FALSE
#' @keywords base-n
#' @export
#' @examples
#' find_base_n(dataset_1, .05, TRUE)
#' find_base_n(dataset_2, .001)

find_base_n <- function(data_set, alpha=.05, paired=FALSE) {

#initialize pvalues variable as vector
pvalues <- data.frame("participants"= 1, "p.value" = 1);

#Convert piped in dataset to a data frame
data_set <- as.data.frame(
	data_set,
	row.names = NULL,
	optional = FALSE,
	cut.names = FALSE,
	col.names = names(data_set),
	fix.empty.names = TRUE,
	stringsAsFactors = FALSE);


#######Incremental t-Tests------------------------
#i.e. 1-2, 1-3, 1-4, 1-5, etc..
	for(k in 2:nrow(data_set)) {

	#Paired Sample = FALSE - Run independant Sample
		if (paired == FALSE || missing(paired) == TRUE) {

			#Makes sure there is variance prior to running ind-sample t-test
			#Use apply to get variance values rows 1:k, in columns 1 and 2
			#Use all to compare variance to 0
			if (all(apply(data_set[1:k,], 2, var) != 0 )){

				#Saves iterative p values in a vector
				pvalues <- rbind(pvalues, c(k, t.test(data_set[1:k,1],data_set[1:k,2])$p.value));
			}
			else {
				#if no variance found, return -1
				return(-1);
			}
		} #end if


	#Paired Sample = TRUE - Run paired Sample
		if (paired == TRUE) {

			#Makes sure there is variance before running paired sample t-test
			#Use apply to get variance values in both rows and colums
			#Use all to compare variances to 0
			if (sum(apply(data_set[1:k,], 1, var)) > 0 &
				sum(apply(data_set[1:k,], 2, var)) > 0 ){

				#tryCatch---
				tryCatch({ #checks for errors
				#Saves iterative p values in a vector
				pvalues <- rbind(pvalues, c(k, t.test(data_set[1:k,1],data_set[1:k,2], paired = TRUE)$p.value));
				},

				#If there is an error, return -1
				error = function(err) {
					return(-1);
				}); #End tryCatch---
			}

			#If there is not variance, return -1
			else {
				return(-1);
			}
		} #end if
} #End For Loop

	i=2; #starts with atleast 2 participants

######Minimum Sig. Value------------------

	#For debugging, saves list of pvalues to Global env
	assign('pvalues', pvalues, envir=.GlobalEnv);

	###Minimum base_n allowed----------
	min_n <- 10;

	#Return number of participants needed for significance with a minimum number of participants examined
	min_p = head(pvalues[pvalues$p.value <= alpha & pvalues$participants >= min_n,], 1);

	if (length(min_p$participants) > 0) {
		return(min_p$participants)
	} else {
		return(0)
		}
}