Added gini score implementation to checkpoint1

checkpoint1/README.md

@@ -0,0 +1 @@
+Please goto the following link https://docs.google.com/a/springer.com/presentation/d/1cC3opKIVAHYFXjWB1sTE5FO7aii9xPBeOw5qC18puuU/edit?usp=sharing

checkpoint1/src/main/kotlin/com/springernature/ctree4k/Feature.kt

@@ -0,0 +1,25 @@
+package com.springernature.ctree4k
+
+data class FeatureId(val value: String)
+data class Feature(val value: Double, val id: FeatureId)
+
+data class InstanceClass(val value: Int, val name: String)
+data class InstanceClasses(val value: Set<InstanceClass>)
+
+interface Instance {
+
+    val features: List<Feature>
+
+    val instanceClass: InstanceClass
+
+    fun get(featureId: FeatureId): Feature {
+        val feature = features.find { it.id == featureId }
+
+        if(feature != null) {
+            return feature
+        } else {
+
+            throw IllegalArgumentException("Could not find feature in Instance[$this] with id[$featureId]")
+        }
+    }
+}

checkpoint1/src/main/kotlin/com/springernature/ctree4k/GiniImpurityScore.kt

@@ -0,0 +1,42 @@
+package com.springernature.ctree4k
+
+data class SplitHalf(val instances: List<Instance>)
+
+class GiniImpurityScore : (Pair<SplitHalf, SplitHalf>, InstanceClasses) -> GiniScore {
+
+    override fun invoke(bothSplits: Pair<SplitHalf, SplitHalf>, allClasses: InstanceClasses): GiniScore {
+
+        val eachScore = bothSplits.toList().map { half: List<Instance> ->
+            val currentSplitSize = half.size
+
+            val eachClassScoreBySplit = allClasses.value.map { featureSetClass: InstanceClass ->
+                val ratio = half.ratioThatAre(featureSetClass)
+                ratio * ratio
+            }
+
+            val score = eachClassScoreBySplit.sum()
+
+            val scoreBySplit = (1.0 - score) * (currentSplitSize.toDouble().div(bothSplits.count()))
+            scoreBySplit
+        }
+        val score = eachScore.sum()
+
+        return GiniScore(score)
+    }
+}
+
+data class GiniScore(val value: Double)
+
+private fun List<Instance>.ratioThatAre(instanceClass: InstanceClass): Double {
+    val count = count {
+        it.instanceClass == instanceClass
+    }
+
+    return count.toDouble().div(size.toDouble())
+}
+
+private fun Pair<SplitHalf, SplitHalf>.toList(): List<List<Instance>> {
+    return listOf(first.instances, second.instances)
+}
+
+private fun Pair<SplitHalf, SplitHalf>.count(): Int = first.instances.count().plus(second.instances.count())

checkpoint1/src/test/kotlin/com/springernature/ctree4k/GiniImpurityScoreTest.kt

@@ -0,0 +1,52 @@
+package com.springernature.ctree4k
+
+import com.natpryce.hamkrest.assertion.assertThat
+import com.natpryce.hamkrest.equalTo
+import org.junit.Test
+
+class GiniImpurityScoreTest {
+
+    val triangleClass = InstanceClass(0, "triangle")
+    val circleClass = InstanceClass(1, "circle")
+    val allClasses = InstanceClasses(setOf(triangleClass, circleClass))
+
+    val triangle = createInstance(triangleClass)
+    val circle = createInstance(circleClass)
+
+    val giniScore = GiniImpurityScore()
+
+    @Test fun `returns 0,5`() {
+
+        val leftSplit = SplitHalf(listOf(triangle, circle))
+        val rightSplit = SplitHalf(listOf(triangle, circle))
+
+        assertThat(giniScore(Pair(leftSplit, rightSplit), allClasses), equalTo(GiniScore(0.5)))
+    }
+
+    @Test fun `returns 0`() {
+
+        val leftSplit = SplitHalf(listOf(triangle, triangle))
+        val rightSplit = SplitHalf(listOf(circle, circle))
+
+        assertThat(giniScore(Pair(leftSplit, rightSplit), allClasses), equalTo(GiniScore(0.0)))
+    }
+
+    @Test fun `returns 0,4`() {
+
+        val leftSplit = SplitHalf(listOf(triangle, triangle))
+        val rightSplit = SplitHalf(listOf(circle, circle, circle, circle, triangle, triangle, triangle, triangle))
+
+        assertThat(giniScore(Pair(leftSplit, rightSplit), allClasses), equalTo(GiniScore(0.4)))
+    }
+}
+
+private fun createInstance(instanceClass: InstanceClass): Instance {
+    return object : Instance {
+
+        override val features: List<Feature>
+            get() = emptyList()
+
+        override val instanceClass: InstanceClass
+            get() = instanceClass
+    }
+}

settings.gradle

@@ -1,3 +1,4 @@
 rootProject.name = 'ctree4k'
 include 'checkpoint0'
+include 'checkpoint1'