k-means method on Haskell

kmeans.hs

import Data.List
import System.IO
import System.Random

type Cluster = Int
type Point = (Double, Double)  -- Point in R^2
type CPoint = (Cluster, Point) -- Classified Point

---- Euclid Distance between Two Points
dist :: Point -> Point -> Double
dist (x0,y0) (x1,y1) = sqrt $ (x0-x1)^2 + (y0-y1)^2

---- Center of Points
center :: [Point] -> Point
center = (\(ys,zs) -> (average ys, average zs)).unzip
  where average xs = (sum xs)/((fromIntegral.length) xs)

---- CPoints are Same Cluster or Not
colleague :: CPoint -> CPoint -> Bool
colleague (c0,_) (c1,_) = c0 == c1

--- One Step of K-means Method
kmeans :: [CPoint] -> [CPoint]
kmeans xs = map (\(_,p) -> (classify p, p)) xs
  where
  ---- Classify a Point 
  classify :: Point -> Cluster
  classify x = (snd.last.sort) $ map (\(c,p) -> (dist x p, c)) (centers xs)
  ----
  ---- Calculate Prototypes
  centers :: [CPoint] -> [CPoint]
  centers xs = map ((\(cs,ps) -> (head cs, center ps)).unzip) $ (groupBy colleague).sort $ xs

main = do
  gen <- getStdGen

  handle <- openFile "data.csv" ReadMode
  content <- hGetContents handle

  let
    points :: [Point]
    points = map makePoint $ lines content
      where
      ---- "Double, Double" -> (Double, Double)
      makePoint :: String -> Point
      makePoint ps =
        let substr = groupBy (\x y -> and [x/=',', y/=',']) ps
        in (read $ head substr, read $ last substr)
    ----
    ---- Create Initial Value
    iv :: [CPoint]
    iv = zip (randomRs (0,2) gen :: [Int]) points

    ---- K-means Method !!
    experiment = iterate kmeans iv
    result = (groupBy colleague).sort $ experiment!!100

    output :: Int -> IO()
    output n = do
      handle <- openFile ("result"++(show n)++".csv") WriteMode
      mapM_ (hPutStrLn handle) $ map form (result!!n)
      hClose handle
        where
        form :: CPoint -> String
        form (c,p) = (show $ fst p) ++ ", " ++ (show $ snd p)

  mapM_ output [0..2]

  hClose handle