FR: X2contingency

src/FSharp.Stats/Testing/ChiSquareTest.fs

@@ -36,7 +36,7 @@ module ChiSquareTest =
     open System
     open FSharp.Stats
 
-    /// Computes the Chi-Square test
+    /// Computes the Chi-Square test.
     /// n data points -> degrees of freedom = n - 1 
     let compute (degreesOfFreedom:int) (expected:seq<float>) (observed:seq<float>) =
         //let chechParams =
@@ -53,4 +53,16 @@ module ChiSquareTest =
         TestStatistics.createChiSquare chi2 (float degreesOfFreedom)
 
 
-
+    // Adapted from https://www.ling.upenn.edu/~clight/chisquared.htm
+    /// Computes a Chi-Square test for a 2 × 2 contingency table. Aka Cochran-Mantel-Haenszel (CMH) test, ger. Chi-Quadrat-Vierfeldertest.
+    let contingencyTable2x2Test trait1GroupA trait2GroupA trait1GroupB trait2GroupB =
+        let columnTrait1 = trait1GroupA + trait1GroupB
+        let columnTrait2 = trait2GroupA + trait2GroupB
+        let rowGroupA = trait1GroupA + trait2GroupA
+        let rowGroupB = trait1GroupB + trait2GroupB
+        let total = columnTrait1 + columnTrait2
+        let expFreqA1 = (rowGroupA * columnTrait1) / total
+        let expFreqA2 = (rowGroupA * columnTrait2) / total
+        let expFreqB1 = (rowGroupB * columnTrait1) / total
+        let expFreqB2 = (rowGroupB * columnTrait2) / total
+        compute 1 [expFreqA1; expFreqA2; expFreqB1; expFreqB2] [trait1GroupA; trait2GroupA; trait1GroupB; trait2GroupB]

tests/FSharp.Stats.Tests/Testing.fs

@@ -256,14 +256,35 @@ let chiSquaredTests =
         let observed = [315.;101.;80.;32.;50.] 
         let df = expected.Length - 1
         ChiSquareTest.compute df expected observed
+
+    // taken from: https://de.wikipedia.org/wiki/Chi-Quadrat-Test#Beispiele_und_Anwendungen
+    let testCaseContingency1 =
+        let trait1GroupA = 25.
+        let trait2GroupA = 25.
+        let trait1GroupB = 30.
+        let trait2GroupB = 20.
+        ChiSquareTest.contingencyTable2x2Test trait1GroupA trait2GroupA trait1GroupB trait2GroupB
+
+    let testCaseContingency2 =
+        let trait1GroupA = 250.
+        let trait2GroupA = 250.
+        let trait1GroupB = 300.
+        let trait2GroupB = 200.
+        ChiSquareTest.contingencyTable2x2Test trait1GroupA trait2GroupA trait1GroupB trait2GroupB
 
     testList "Testing.ChiSquaredTest" [
         testCase "compute" <| fun () -> 
             Expect.isTrue (0.9254 = Math.Round(testCase1.PValueRight,4)) "pValue should be equal."
             Expect.isTrue (0.4700 = Math.Round(testCase1.Statistic,4)) "statistic should be equal."
             Expect.isTrue (0.000638 = Math.Round(testCase2.PValueRight,6)) "pValue should be equal."
             Expect.isTrue (19.461 = Math.Round(testCase2.Statistic,3)) "statistic should be equal."
-
+
+        testCase "contingencyTable" <| fun () ->
+            Expect.floatClose Accuracy.veryHigh (round 4 testCaseContingency1.PValueRight) 0.3149 "pValue is not about equal"
+            Expect.floatClose Accuracy.veryHigh (round 3 testCaseContingency1.Statistic) 1.01 "teststatistic is not about equal"
+
+            Expect.floatClose Accuracy.veryHigh (round 4 testCaseContingency2.PValueRight) 0.0015 "pValue is not about equal"
+            Expect.floatClose Accuracy.veryHigh (round 2 testCaseContingency2.Statistic) 10.1 "teststatistic is not about equal"
     ]
 
 [<Tests>]