based off http://www.hakkalabs.co/articles/digging-into-the-dirichlet-distribution

chineseRestaurant.hs

import System.Random (getStdGen, randomR)
import Data.MultiSet (fromList, toOccurList)
import Data.List (maximum)
import Data.Function (on)
import Data.List (sortBy)
import Graphics.Google.Chart
import Hledger.Cli.Utils (openBrowserOn)

step xs g = (newItem : xs, newG)
  where (i, newG) = pick (length xs) g
        pick n = randomR (0, n-1)
        newItem = if xs !! i == 0 
                  then 1 + maximum xs 
                  else xs !! i

steps xs _ 0 = xs
steps xs g n = steps newXs newG (n-1)
  where (newXs, newG) = step xs g

main = do
  g <- getStdGen
  let xs = steps [0,1,2] g 5000
  let freqs = sortBy (compare `on` snd) $ toOccurList $ fromList xs
  putStrLn $ "There are " ++ (show.length) freqs ++ " items"
  print freqs
  openBrowserOn $ piChart freqs

-- limited to data values up to 4095
piChart :: [(Int, Int)] -> String
piChart freqs = chartURL $
  setSize 500 400 $
  setTitle ("Frequency Plot of " ++ show (length freqs) ++ " Items") $
  setData (encodeDataExtended [freqVals]) $
  setLabels freqIds $
  newPieChart Pie2D
  where freqIds = map (show.fst) freqs
        freqVals = map snd freqs