Refactor the code

build.sbt

@@ -33,14 +33,15 @@ scalacOptions := Seq(
 
 scalacOptions in(Compile, doc) := Seq("-no-link-warnings")
 
-scalaVersion := "2.12.18"
+scalaVersion := "2.13.14"
 
-crossScalaVersions := Seq("2.12.18", "2.11.12", "2.13.14", "3.1.3")
+crossScalaVersions := Seq("2.13.14", "2.11.12", "2.12.18", "3.1.3")
 
 libraryDependencies ++= {
   Seq(
+    "org.apache.commons" % "commons-lang3" % "3.15.0",
     "org.apache.commons" % "commons-math3" % "3.6.1",
-    "org.apache.commons" % "commons-text" % "1.10.0",
+    "org.apache.commons" % "commons-text" % "1.12.0",
     "io.spray" %% "spray-json" % "1.3.6",
     "org.scalatest" %% "scalatest" % "3.2.15" % "test",
     "junit" % "junit" % "4.13.2" % "test"

src/main/scala/org/pmml4s/common/DataType.scala

@@ -15,6 +15,7 @@
  */
 package org.pmml4s.common
 
+import org.pmml4s.data.DataVal
 import org.pmml4s.util.Utils
 import org.pmml4s.xml.ValTags
 
@@ -24,7 +25,7 @@ import org.pmml4s.xml.ValTags
 sealed abstract class DataType extends DataTypeLike {
   def dataType: DataType = this
 
-  def toVal(s: String): Any = Utils.toVal(s, this)
+  def toVal(s: String): DataVal = Utils.toDataVal(s, this)
 }
 
 /**
@@ -140,7 +141,7 @@ case object DateTimeType extends DateTimeType
  * since midnight. The time 00:00 is represented by the number 0. No negative values are allowed.
  */
 class TimeSecondsType private() extends TimeType {
-  override def toString = ValTags.TIMESECONDS
+  override def toString: String = ValTags.TIMESECONDS
 }
 
 case object TimeSecondsType extends TimeSecondsType
@@ -156,10 +157,10 @@ case class DateDaySinceYearType(aYear: Int) extends DateType {
 }
 
 case object DateDaySinceYearType {
-  val DateDaySinceYear0Type = DateDaySinceYearType(0)
-  val DateDaySinceYear1960Type = DateDaySinceYearType(1960)
-  val DateDaySinceYear1970Type = DateDaySinceYearType(1970)
-  val DateDaySinceYear1980Type = DateDaySinceYearType(1980)
+  val DateDaySinceYear0Type: DateDaySinceYearType = DateDaySinceYearType(0)
+  val DateDaySinceYear1960Type: DateDaySinceYearType = DateDaySinceYearType(1960)
+  val DateDaySinceYear1970Type: DateDaySinceYearType = DateDaySinceYearType(1970)
+  val DateDaySinceYear1980Type: DateDaySinceYearType = DateDaySinceYearType(1980)
 }
 
 /**
@@ -174,10 +175,10 @@ case class DateTimeSecondSinceYearType(aYear: Int) extends DateTimeType {
 }
 
 case object DateTimeSecondSinceYearType {
-  val DateTimeSecondSinceYear0Type = DateTimeSecondSinceYearType(0)
-  val DateTimeSecondSinceYear1960Type = DateTimeSecondSinceYearType(1960)
-  val DateTimeSecondSinceYear1970Type = DateTimeSecondSinceYearType(1970)
-  val DateTimeSecondSinceYear1980Type = DateTimeSecondSinceYearType(1980)
+  val DateTimeSecondSinceYear0Type: DateTimeSecondSinceYearType = DateTimeSecondSinceYearType(0)
+  val DateTimeSecondSinceYear1960Type: DateTimeSecondSinceYearType = DateTimeSecondSinceYearType(1960)
+  val DateTimeSecondSinceYear1970Type: DateTimeSecondSinceYearType = DateTimeSecondSinceYearType(1970)
+  val DateTimeSecondSinceYear1980Type: DateTimeSecondSinceYearType = DateTimeSecondSinceYearType(1980)
 }
 
 /**

src/main/scala/org/pmml4s/common/ScoreDistribution.scala

@@ -15,11 +15,13 @@
  */
 package org.pmml4s.common
 
+import org.pmml4s.data.DataVal
+
 /**
  * Comprises a method to list predicted values in a classification trees structure.
  */
 class ScoreDistribution(
-                         val value: Any,
+                         val value: DataVal,
                          val recordCount: Double,
                          val confidence: Option[Double],
                          val probability: Option[Double]) extends PmmlElement {
@@ -34,19 +36,15 @@ class ScoreDistribution(
 trait HasScoreDistributions {
   def scoreDistributions: ScoreDistributions
 
-  def getConfidence(value: Any): Double = scoreDistributions.valueToDistribution.get(value).flatMap(_.confidence) getOrElse
-    Double.NaN
-
-  def getProbability(value: Any): Double = scoreDistributions.valueToDistribution.get(value).flatMap(_.probability) getOrElse
-    Double.NaN
+  def getConfidence(value: DataVal): Double = scoreDistributions.confidences.getOrElse(value, Double.NaN)
 
-  def getScoreDistribution(value: Any): Option[ScoreDistribution] = scoreDistributions.valueToDistribution.get(value)
+  def getProbability(value: DataVal): Double = scoreDistributions.probabilities.getOrElse(value, Double.NaN)
 
-  def probabilities: Map[Any, Double] = scoreDistributions.valueToDistribution.map(x => (x._1, x._2.probability.getOrElse(Double.NaN))).toMap
+  def probabilities: Map[DataVal, Double] = scoreDistributions.probabilities
 }
 
 class ScoreDistributions(val scoreDistributions: Array[ScoreDistribution]) extends PmmlElement {
-  val valueToDistribution: Map[Any, ScoreDistribution] = {
+  val valueToDistribution: Map[DataVal, ScoreDistribution] = {
     val total = scoreDistributions.map(_.recordCount).sum
 
     // Compute probabilities if not present.
@@ -59,9 +57,13 @@ class ScoreDistributions(val scoreDistributions: Array[ScoreDistribution]) exten
     } else scoreDistributions).map(x => (x.value, x)).toMap
   }
 
+  val probabilities: Map[DataVal, Double] = valueToDistribution.map(x => (x._1, x._2.probability.getOrElse(Double.NaN)))
+
+  val confidences: Map[DataVal, Double] = valueToDistribution.map(x => (x._1, x._2.confidence.getOrElse(Double.NaN)))
+
   def this() = {
     this(Array.empty)
   }
 
-  def classes: Array[Any] = scoreDistributions.map(_.value)
+  def classes: Array[DataVal] = scoreDistributions.map(_.value)
 }

src/main/scala/org/pmml4s/common/Table.scala

@@ -15,45 +15,46 @@
  */
 package org.pmml4s.common
 
+import org.pmml4s.data.DataVal
 import org.pmml4s.xml.ElemTags.{INLINE_TABLE, TABLE_LOCATOR}
 
 sealed trait Table extends PmmlElement {
-  def find(inputs: Map[String, Any], output: String): Option[Any]
+  def find(inputs: Map[String, DataVal], output: String): Option[DataVal]
 
   def apply(i: Int): Row
 
   def dim: (Int, Int)
 }
 
 object Table {
-  val values = Set(TABLE_LOCATOR, INLINE_TABLE)
+  val values: Set[String] = Set(TABLE_LOCATOR, INLINE_TABLE)
 
-  def contains(s: String) = values.contains(s)
+  def contains(s: String): Boolean = values.contains(s)
 }
 
 class InlineTable(val rows: Array[Row]) extends Table {
-  override def find(inputs: Map[String, Any], output: String): Option[Any] = {
-    val r = rows.find(x => inputs.forall(p => x.elements.get(p._1) == Some(p._2)))
+  override def find(inputs: Map[String, DataVal], output: String): Option[DataVal] = {
+    val r = rows.find(x => inputs.forall(p => x.elements.get(p._1).contains(p._2)))
     if (r.isDefined) r.get.elements.get(output) else None
   }
 
   override def apply(i: Int): Row = rows(i)
 
-  override def dim: (Int, Int) = (rows.size, if (rows.nonEmpty) rows.head.size else 0)
+  override def dim: (Int, Int) = (rows.length, if (rows.nonEmpty) rows.head.size else 0)
 }
 
 class TableLocator extends Table {
-  override def find(inputs: Map[String, Any], output: String): Option[Any] = ???
+  override def find(inputs: Map[String, DataVal], output: String): Option[DataVal] = ???
 
   override def apply(i: Int): Row = ???
 
   override def dim: (Int, Int) = ???
 }
 
-class Row(val elements: Map[String, Any]) extends PmmlElement {
+class Row(val elements: Map[String, DataVal]) extends PmmlElement {
   def size: Int = elements.size
 
-  def apply(name: String): Any = elements(name)
+  def apply(name: String): DataVal = elements(name)
 
-  def get(name: String): Option[Any] = elements.get(name)
+  def get(name: String): Option[DataVal] = elements.get(name)
 }

src/main/scala/org/pmml4s/common/Value.scala

@@ -16,6 +16,7 @@
 package org.pmml4s.common
 
 import org.pmml4s.common.Property.Property
+import org.pmml4s.data.DataVal
 import org.pmml4s.util.Utils
 
 import scala.collection.mutable
@@ -26,15 +27,15 @@ class Value(
              val property: Property = Property.valid) extends PmmlElement
 
 object Value {
-  def distinguish(values: Array[Value], dataType: DataType): (Array[Any], Set[Any], Set[Any], Map[Any, String]) = {
-    val valid = mutable.ArrayBuilder.make[Any]
+  def distinguish(values: Array[Value], dataType: DataType): (Array[DataVal], Set[DataVal], Set[DataVal], Map[DataVal, String]) = {
+    val valid = mutable.ArrayBuilder.make[DataVal]
     valid.sizeHint(values)
-    val invalid = new mutable.HashSet[Any]()
-    val missing = new mutable.HashSet[Any]()
-    val labels = new mutable.HashMap[Any, String]()
+    val invalid = new mutable.HashSet[DataVal]()
+    val missing = new mutable.HashSet[DataVal]()
+    val labels = new mutable.HashMap[DataVal, String]()
 
     for (v <- values) {
-      val a = Utils.toVal(v.value, dataType)
+      val a = Utils.toDataVal(v.value, dataType)
       v.property match {
         case Property.valid   => valid += a
         case Property.invalid => invalid += a
@@ -67,6 +68,6 @@ object Value {
  * explicitly define values which are interpreted as missing values.
  */
 object Property extends Enumeration {
-  type Property = Value
-  val valid, invalid, missing = Value
+  type Property = this.Value
+  val valid, invalid, missing = this.Value
 }

src/main/scala/org/pmml4s/common/Vector.scala

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2017-2019 AutoDeploy AI
+ * Copyright (c) 2017-2024 AutoDeploy AI
  *
  * Licensed under the Apache License, Version 2.0 (the "License");
  * you may not use this file except in compliance with the License.
@@ -27,7 +27,7 @@ trait Vector[@specialized(Int, Long, Float, Double) V] extends java.io.Serializa
 
   def length: Int = size
 
-  def toArray(implicit cm: ClassTag[V]) = {
+  def toArray(implicit cm: ClassTag[V]): Array[V] = {
     val result = new Array[V](length)
     var i = 0
     while (i < length) {

src/main/scala/org/pmml4s/common/elements.scala

@@ -17,6 +17,7 @@ package org.pmml4s.common
 
 import org.pmml4s.common.MiningFunction.MiningFunction
 import org.pmml4s.model.Model
+import org.pmml4s.util.StringUtils
 
 import scala.collection.immutable
 
@@ -27,7 +28,7 @@ trait HasVersion {
   def version: String = parent.version
 
   /** Returns PMML version as a double value */
-  def dVersion: Double = version.toDouble
+  def dVersion: Double = StringUtils.asDouble(version)
 }
 
 case class Extension(extender: Option[String], name: Option[String], value: Option[Any], content: Option[Any])
@@ -57,7 +58,7 @@ trait PmmlElement extends HasExtensions with Serializable
 class Application(
                    val name: String,
                    val version: Option[String]) extends PmmlElement {
-  override def toString = if (version.isDefined) (name + " " + version.get) else name
+  override def toString: String = if (version.isDefined) (name + " " + version.get) else name
 }
 
 /**