Systematic Sampling in R

SampleBy in the doBy package that was useful in R is not supported.

I have created sample.by systematic sampling function as below. ^^

## usage : sample.by(data_as_dataframe, number_of_column, ratio_of_sample)

##             returns a list of sample.df and rest.df

 

sample.by <- function(df, by.col.loc=1, prop=0.1) {

  sample.df <- data.frame()

  rest.df <- data.frame()

  dat <- split(df, df[by.col.loc])

  for(i in 1:NROW(dat)) {

    idx <- sample(c(rep(1,n<-round(NROW(dat[[i]]) * prop)),

                    rep(2,NROW(dat[[i]]) - n)))

    sample.df <- rbind(sample.df, dat[[i]][idx==1,])

    rest.df  <- rbind(rest.df, dat[[i]][idx==2,])    

  }

  list(sample.df=sample.df[sample(NROW(sample.df)),],

       rest.df=rest.df[sample(NROW(rest.df)),])

}

## example

sample.by(iris, 5, 0.7)