MetricBySplit produces a data.frame describing one metric variable by an optional SPLIT grouping

The function produces a data.frame describing one metric variable METRIC with various statistics SPLIT is a group variable which may be left out. Tests may be preformed on Mean (t.test, oneway.test) or Median (wilcox.test,kruskal.test)

MetricBySplit(METRIC, SPLIT = NULL, VARNAME = T, LONGNAME = NULL,
  PVALUE = F, TEST = NULL, ALL = T, SEL = NULL, SHOW_N = T,
  DIGITS = 2)

Arguments

METRIC	is a numeric variable
SPLIT	=NULL split factor (group variable), if NULL an ungrouped description is returned
VARNAME	=T indicate the VARNAME name of METRICS in the first column
LONGNAME	=NULL optional long name for METRICS if NULL a LONGNAME is derived from the call
PVALUE	=F If TRUE, tests on mean and Median are performed and added as last column
TEST	=NULL Choose TEST from c("Mean","Median"). If not specified set to both i.e. c("Mean","Median")
ALL	=T If TRUE, also overall statistics
SEL	=NULL if specified select statistics from c("Mean", "Sd", "Median", "Mad", "Perc25", "Perc75", "Min", "Max")
SHOW_N	=T Show N=length(METRIC) top left
DIGITS	=2 Number of digits for statistics

Value

a data.frame OF CHARACTERS with statistics by the group variable SPLIT

Details

MetricBySplit

Examples

if (FALSE) {
Metric<-c(rnorm(57),rnorm(43,2,2))
Split<-c(rep("Arm A",57),rep("Arm B",43))
# 
MetricBySplit(Metric,Split)
MetricBySplit(Metric,Split,PVALUE=T,TEST=c("Mean","Median"),
                           SEL=c("Mean","Median"),DIGITS=3)
}