tests/Clustering/FrequentItemSet/BruteForceSpec.hs

   1 {-# LANGUAGE OverloadedLists #-}
   2 {-# LANGUAGE OverloadedStrings #-}
   3
   4 module Clustering.FrequentItemSet.BruteForceSpec where
   5
   6 import Control.Monad (forM_, return)
   7 import Data.Function (on, ($), (&), (.))
   8 import Data.Functor ((<$>), (<&>))
   9 import Data.Int (Int)
  10 import Data.List qualified as List
  11 import Data.Ord (Ord (..))
  12 import Data.Ratio (Rational, (%))
  13 import Data.Semigroup (Semigroup (..))
  14 import Data.Sequence qualified as Seq
  15 import Data.Set qualified as Set
  16 import Data.Text.Short qualified as TextS
  17 import GHC.IsList (toList)
  18 import Logic
  19 import Numeric.Probability (assertProbability)
  20 import System.FilePath ((<.>))
  21 import Test.Syd
  22 import Text.Show (Show (..))
  23 import Prelude (Num)
  24
  25 import Clustering.FrequentItemSet.BruteForce
  26 import Utils
  27
  28 -- | From https://research.nii.ac.jp/~uno/code/lcm.html#IntroductionstoFrequentItemsetMining
  29 databaseTakeakiUno :: Ord item => Num item => [Transaction item ()]
  30 databaseTakeakiUno =
  31   Transaction ()
  32     <$> [ [1, 2, 5, 6, 7]
  33         , [2, 3, 4, 5]
  34         , [1, 2, 7, 8, 9]
  35         , [1, 7, 9]
  36         , [2, 7, 9]
  37         , [2, 7, 9] -- Copy-paste typo on the original example
  38         , [1, 9] -- Add this to increase the support of [1,9] because the original example is wrong…
  39         , [2]
  40         ]
  41
  42 -- | From https://hal.science/hal-03500847
  43 databaseHAL03500847T2 :: Ord item => Num item => [Transaction item ()]
  44 databaseHAL03500847T2 =
  45   Transaction ()
  46     <$> [ [1, 3, 7, 6]
  47         , [1, 2, 7]
  48         , [2, 8, 9, 10]
  49         , [5, 11]
  50         ]
  51
  52 database1 :: [Transaction TextS.ShortText ()]
  53 database1 =
  54   Transaction ()
  55     <$> [["a", "b"]]
  56
  57 spec :: Spec
  58 spec = do
  59   {-
  60   describe "allFrequentItemSets" do
  61     forM_ ([2 .. 3] :: [Int]) \minSupp ->
  62       goldenShow ("db=TakeakiUno" <.> "minSupp=" <> show minSupp) $
  63         allFrequentItemSets @Int
  64           (unitName databaseTakeakiUno)
  65           (assertStrictlyPositive minSupp)
  66     forM_ ([2 .. 3] :: [Int]) \minSupp ->
  67       goldenShow ("db=HAL03500847T2" <.> "minSupp=" <> show minSupp) $
  68         allFrequentItemSets @Int
  69           (unitName databaseHAL03500847T2)
  70           (assertStrictlyPositive minSupp)
  71
  72   describe "associationRules" do
  73     forM_ ([2 .. 3] :: [Int]) \minSupp ->
  74       forM_ ([(75 % 100)] :: [Rational]) \minConf ->
  75         goldenShow
  76           ("db=TakeakiUno" <.> "minSupp=" <> show minSupp <.> "minConf=75%")
  77           [ associationRules
  78             fis
  79             (unitName databaseTakeakiUno)
  80             (unitName (assertProbability minConf))
  81           | fis <-
  82               allFrequentItemSets @Int
  83                 (unitName databaseTakeakiUno)
  84                 (assertStrictlyPositive minSupp)
  85           ]
  86   describe "allClosedFrequentItemSets" do
  87     forM_ ([2 .. 3] :: [Int]) \minSupp ->
  88       forM_ ([2 .. 3] :: [Int]) \minSize ->
  89         goldenShow ("db=TakeakiUno" <.> "minSupp=" <> show minSupp <.> "minSize=" <> show minSize) $
  90           allClosedFrequentItemSets @Int
  91             (assertStrictlyPositive minSupp)
  92             (assertStrictlyPositive minSize)
  93             (unitName (databaseTakeakiUno @Int))
  94             & unName
  95     forM_ ([1 .. 3] :: [Int]) \minSupp ->
  96       forM_ ([2 .. 4] :: [Int]) \minSize ->
  97         goldenShow ("db=HAL03500847T2" <.> "minSupp=" <> show minSupp <.> "minSize=" <> show minSize) $
  98           allClosedFrequentItemSets @Int
  99             (assertStrictlyPositive minSupp)
 100             (assertStrictlyPositive minSize)
 101             (unitName (databaseHAL03500847T2 @Int))
 102             & unName
 103   describe "allClosedFrequentItemSets" do
 104     forM_ ([1 .. 1] :: [Int]) \minSupp ->
 105       forM_ ([1 .. 1] :: [Int]) \minSize ->
 106         goldenShow ("db=1" <.> "minSupp=" <> show minSupp <.> "minSize=" <> show minSize) $
 107           allClosedFrequentItemSets @TextS.ShortText
 108             (assertStrictlyPositive minSupp)
 109             (assertStrictlyPositive minSize)
 110             (unitName database1)
 111             & unName
 112   -}
 113   return ()