Collection of helpful EDA functions in R (originally created for M.S. in Data Science assignment work at Regis University)

eda.r

hist.density <- function (data, xlab = "<x>", font.main = 1,
                          main = "Histogram of data") {

  # Calculate histogram based on these values
  data.hist = hist(data, plot = FALSE)
  
  # Determine scaling factor
  multiplier = data.hist$counts / data.hist$density
  
  # Create density function of the area
  data.density = density(data)
  
  # Scale y-axis of density to histogram limits
  data.density$y = data.density$y * multiplier[1]
  
  # Plot histgram with labels and limits
  plot(data.hist, xlab = xlab, main = main, font.main = font.main,
       ylim = c(0, max(c(data.density$y, data.hist$counts))))
  
  # Add density function as a line overlay
  lines(data.density)
}

dist.summary <- function (dataframe, vdescriptions = names(dataframe),
                          rows = 3, columns = 3,
                          main = 'Distribution of Variables',
                          mar = c(1, 1, 2, 0), oma = c(1, 1, 3, 1)) {
  # Setup layout and spacing
  total = rows * columns
  layout(mat = matrix(1:(total * 2), rows * 2, columns, byrow = F), height = c(rep(c(2, 3), total)))
  par(mar = mar, oma = oma)
  
  for (index in 1:ncol(dataframe)) {
    column = dataframe[,names(dataframe)[index]]
    
    # Check for discrete / continuous
    if (is.factor(column)) {
      # Default padding
      par(mai = rep(0.3, 4))
      plot.new()
      title(main = vdescriptions[index])
      
      # Padding (except top)
      par(mai = c(0.3, 0.3, 0, 0.3))
      
      # Display bar plot of frequencies
      barplot(table(column))
    } else {
      # Padding (expanded for top and none for bottom)
      par(mai = c(0, 0.3, 0.5, 0.3))
      boxplot(column, cex = 0.8, horizontal = T, pch = '.',
              main = vdescriptions[index], outline = F, axes = F)
      
      # Padding (except top)
      par(mai = c(0.3, 0.3, 0, 0.3))
      
      # Display histogram and density function
      hist.density(column, font.main = 1, main = '')
    }
  }
  
  title(main = main, outer = T)
}