R/split_by.R

In alemiani/explora: Explore and Visualize Data

split_by <- function(x, split = NULL,
                         factors = NULL,
                         split.mode = c('minor', 'minor.equal', 'major.equal')
                         ) {
    g.name <- as.character(substitute(x)[[3]])

    if(!is.null(factors)) {
      if (length(factors) != 2)
        {stop("\n\nYou must name exactly two factors, otherwise let me do it for you ;-)")}
      low.factor <- factors[1]
      high.factor <- factors[2]
    } else {
      low.factor <- as.character(paste('low.', g.name, sep = ''))
      high.factor <- paste('high.', g.name, sep = '')
    }

    name <- NA


    if(is.null(split) || split == 'median') {
      if(missing(split.mode)) {
        name[x <= median(x, na.rm = T)] <- low.factor
        name[x > median(x, na.rm = T)] <- high.factor
      } else
        if(split.mode == 'minor.equal') {
          name[x <= median(x, na.rm = T)] <- low.factor
          name[x > median(x, na.rm = T)] <- high.factor
        } else
          if(split.mode == 'minor') {
            name[x < median(x, na.rm = T)] <- low.factor
            name[x > median(x, na.rm = T)] <- high.factor
          } else
            if(split.mode == 'major.equal') {
              name[x < median(x, na.rm = T)] <- low.factor
              name[x >= median(x, na.rm = T)] <- high.factor
          } else  {stop("\n\nYou should define how the split has to be performed, otherwise let me do it for you ;-)")}
    }

    if(is.numeric(split)) {
      if(missing(split.mode)) {
        name[x <= split] <- low.factor
        name[x > split] <- high.factor
      } else
        if(split.mode == 'minor.equal') {
          name[x <= split] <- low.factor
          name[x > split] <- high.factor
        } else
          if(split.mode == 'minor') {
            name[x < split] <- low.factor
            name[x > split] <- high.factor
          } else
            if(split.mode == 'major.equal') {
              name[x < split] <- low.factor
              name[x >= split] <- high.factor
            } else  {stop("\n\nYou should define how the split has to be performed, otherwise let me do it for you ;-)")}
    }

    if(split == 'mean') {
      if(missing(split.mode)) {
        name[x <= mean(x, na.rm = T)] <- low.factor
        name[x > mean(x, na.rm = T)] <- high.factor
      } else
        if(split.mode == 'minor.equal') {
          name[x <= mean(x, na.rm = T)] <- low.factor
          name[x > mean(x, na.rm = T)] <- high.factor
        } else
          if(split.mode == 'minor') {
            name[x < mean(x, na.rm = T)] <- low.factor
            name[x > mean(x, na.rm = T)] <- high.factor
          } else
            if(split.mode == 'major.equal') {
              name[x < mean(x, na.rm = T)] <- low.factor
              name[x >= mean(x, na.rm = T)] <- high.factor
            } else  {stop("\n\nYou should define how the split has to be performed, otherwise let me do it for you ;-)")}
    }

    if(split == 'fivenum') {
      if(!is.null(factors)) {stop('\n\nLet me select your factor names for you ;-)')}
      if(!missing(split.mode)) {stop("\n\n No way to select 'split.mode' in here")}
      g1 <- paste(g.name, '_1', sep = '')
      g2 <- paste(g.name, '_2', sep = '')
      g3 <- paste(g.name, '_3', sep = '')
      g4 <- paste(g.name, '_4', sep = '')
      name <- symnum(x, legend = F, corr = FALSE, na = FALSE,
                     cutpoints = c(fivenum(x, na.rm = TRUE)[1],
                                   fivenum(x, na.rm = TRUE)[2],
                                   fivenum(x, na.rm = TRUE)[3],
                                   fivenum(x, na.rm = TRUE)[4],
                                   fivenum(x, na.rm = TRUE)[5]),
                     symbols = c(g1, g2, g3, g4))
    }



    name <- as.factor(name)
  return(name)
}
# split_by(mtcars$mpg, 'median')
# split_by(mtcars$mpg, 'mean')
# split_by(mtcars$mpg, 0, factors = c('jhg', 'kjhg'))
# split_by(mtcars$mpg, split = 'fivenum')
# split_by(mtcars$mpg , factors = c('belowMedian', 'aboveMedian'))