Many resources are needed to download a project. Please understand that we have to compensate our server costs. Thank you in advance. Project price only 1 $
You can buy this project and download/modify it how often you want.
/*
* Copyright (c) 2017-2023 AutoDeployAI
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
package org.pmml4s.model
import org.pmml4s.common.MiningFunction.MiningFunction
import org.pmml4s.common.Predication._
import org.pmml4s.common._
import org.pmml4s.data.Series
import org.pmml4s.metadata.{MiningSchema, Output, OutputField, Targets}
import org.pmml4s.transformations.LocalTransformations
import scala.collection.mutable.ArrayBuffer
import scala.collection.{immutable, mutable}
/**
* The TreeModel in PMML allows for defining either a classification or prediction structure. Each Node holds a logical
* predicate expression that defines the rule for choosing the Node or any of the branching Nodes.
*/
class TreeModel(
var parent: Model,
override val attributes: TreeAttributes,
override val miningSchema: MiningSchema,
val node: Node,
override val output: Option[Output] = None,
override val targets: Option[Targets] = None,
override val localTransformations: Option[LocalTransformations] = None,
override val modelStats: Option[ModelStats] = None,
override val modelExplanation: Option[ModelExplanation] = None,
override val modelVerification: Option[ModelVerification] = None,
override val extensions: immutable.Seq[Extension] = immutable.Seq.empty)
extends Model with HasWrappedTreeAttributes {
/** Model element type. */
override def modelElement: ModelElement = ModelElement.TreeModel
// Optimize the ensemble tree model, ignore the data prepare if it's identical to the parent model
private val ignoreDataPrepare = if (isSubModel && localTransformations.isEmpty) {
var ignore = true
var i = 0
val miningSchemaParent = parent.miningSchema
val len = miningSchema.inputMiningFields.length
while (i < len && ignore) {
val mf = miningSchema.inputMiningFields(i)
if (!mf.isDefault) {
val mfParent = miningSchemaParent.get(mf.name)
if (mfParent.isDefined && mf != mfParent.get) {
ignore = false
}
}
i += 1
}
ignore
} else false
/** Predicts values for a given data series using the model loaded. */
override def predict(values: Series): Series = {
val (series, returnInvalid) = if (ignoreDataPrepare) (values, false) else prepare(values)
if (returnInvalid) {
return nullSeries
}
val outputs = createOutputs()
// The root node could be leaf
var finalNode: Option[Node] = if (node.isLeaf) Some(node) else None
var numMissingCount = 0
var selected = node
var done = false
while (!done && selected.isSplit) {
var child: Node = null
var r = Predication.FALSE
var hit = false
var unknown = false
var i = 0
while (i < selected.children.length && !hit) {
val c = selected.children(i)
c.eval(series) match {
case Predication.TRUE => {
r = Predication.TRUE
child = c
hit = true
}
case Predication.SURROGATE => {
r = Predication.SURROGATE
child = c
hit = true
}
case Predication.UNKNOWN => {
unknown = true
}
case _ =>
}
i += 1
}
if (!hit) {
r = if (unknown) Predication.UNKNOWN else Predication.FALSE
}
if (r == Predication.SURROGATE) {
numMissingCount += 1
}
if (r == Predication.UNKNOWN) {
missingValueStrategy match {
case MissingValueStrategy.`lastPrediction` => {
finalNode = Some(selected)
done = true
}
case MissingValueStrategy.`nullPrediction` =>
done = true
case MissingValueStrategy.`defaultChild` => {
child = selected.defaultChildNode.orNull
numMissingCount += 1
}
case MissingValueStrategy.`weightedConfidence` => if (isClassification) {
val total = selected.recordCount.getOrElse(Double.NaN)
val candidates = selected.children.filter { x => x.eval(series) == UNKNOWN }
var max = 0.0
for (cls <- classes) {
var conf = 0.0
for (cand <- candidates) {
conf += cand.getConfidence(cls) * cand.recordCount.getOrElse(0.0) / total
}
if (conf > max) {
max = conf
outputs.predictedValue = cls
outputs.confidence = conf
}
}
done = true
}
case MissingValueStrategy.`aggregateNodes` => if (isClassification) {
val leaves = mutable.HashSet.empty[Node]
traverseLeaves(selected, series, leaves)
if (leaves.nonEmpty) {
val records = new Array[Double](numClasses)
leaves.foreach(x => {
var i = 0
while (i < numClasses) {
x.scoreDistributions.valueToDistribution.get(classes(i)).foreach(y => records(i) += y.recordCount)
i += 1
}
})
var max = 0.0
var i = 0
while (i < numClasses) {
if (records(i) > max) {
max = records(i)
outputs.predictedValue = classes(i)
outputs.confidence = max / records.sum
}
i += 1
}
}
done = true
}
case MissingValueStrategy.`none` =>
}
}
// Handling the situation where scoring cannot continue
if (child == null && outputs.predictedValue == null) {
noTrueChildStrategy match {
case NoTrueChildStrategy.`returnNullPrediction` => done = true
case NoTrueChildStrategy.`returnLastPrediction` => {
finalNode = Some(selected)
done = true
}
}
} else if (child != null) {
selected = child
if (selected.isLeaf) {
finalNode = Some(selected)
}
}
}
if (finalNode.isDefined) {
selected = finalNode.get
outputs.entityId = selected.id.orNull
outputs.predictedValue = selected.score.orNull
if (isClassification) {
outputs.confidence = selected.getConfidence(outputs.predictedValue)
if (numMissingCount > 0 && missingValuePenalty != 1.0) {
outputs.confidence *= Math.pow(missingValuePenalty, numMissingCount)
}
classes.foreach(x => outputs.putProbability(x, selected.getProbability(x)))
}
}
result(series, outputs)
}
/** The sub-classes can override this method to provide classes of target inside model. */
override def inferClasses: Array[Any] = {
firstLeaf.scoreDistributions.classes
}
/** Returns all candidates output fields of this model when there is no output specified explicitly. */
override lazy val defaultOutputFields: Array[OutputField] = {
val result = mutable.ArrayBuilder.make[OutputField]
result += OutputField.predictedValue(this)
// Check if the first leaf node contains score distributions
val leaf = firstLeaf
if (isClassification && leaf.scoreDistributions.valueToDistribution.nonEmpty) {
if (leaf.scoreDistributions.valueToDistribution.head._2.confidence.isDefined) {
result += OutputField.confidence()
}
if (leaf.scoreDistributions.valueToDistribution.head._2.probability.isDefined) {
result += OutputField.probability()
for (cls <- classes) {
result += OutputField.probability(cls)
}
}
}
if (leaf.id.isDefined) {
result += OutputField.nodeId()
}
result.result()
}
private def firstLeaf: Node = {
var n = node;
while (n.isSplit) n = n.children(0);
n
}
// The method is not used anymore, it's moved into predict directly for performance
private def traverseNode(n: Node, series: Series): (Option[Node], Predication) = {
var unknown = false
// The performance of while loop is better than for comprehension
var i = 0
while (i < n.children.length) {
val child = n.children(i)
child.eval(series) match {
case TRUE => return (Some(child), TRUE)
case SURROGATE => return (Some(child), SURROGATE)
case UNKNOWN => unknown = true
case _ =>
}
i += 1
}
(None, if (unknown) UNKNOWN else FALSE)
}
private def traverseLeaves(n: Node, series: Series, leaves: mutable.Set[Node]): Unit = {
if (n.isLeaf) {
leaves += n
} else {
val candidates = ArrayBuffer.empty[Node]
var i = 0
var done = false
while (!done && i < n.size) {
n(i).eval(series) match {
case TRUE => candidates += n(i); done = true
case SURROGATE => candidates += n(i); done = true
case UNKNOWN => candidates += n(i)
case _ =>
}
i += 1
}
for (child <- candidates)
traverseLeaves(child, series, leaves)
}
}
override def createOutputs(): TreeOutputs = new TreeOutputs
}
/**
* Defines a strategy for dealing with missing values.
*/
object MissingValueStrategy extends Enumeration {
type MissingValueStrategy = Value
/**
* - lastPrediction: If a Node's predicate evaluates to UNKNOWN while traversing the tree, evaluation is stopped
* and the current winner is returned as the final prediction.
* - nullPrediction: If a Node's predicate value evaluates to UNKNOWN while traversing the tree, abort the scoring
* process and give no prediction.
* - defaultChild: If a Node's predicate value evaluates to UNKNOWN while traversing the tree, evaluate the
* attribute defaultChild which gives the child to continue traversing with. Requires the presence of the attribute
* defaultChild in every non-leaf Node.
* - weightedConfidence: If a Node's predicate value evaluates to UNKNOWN while traversing the tree, the
* confidences for each class is calculated from scoring it and each of its sibling Nodes in turn (excluding any
* siblings whose predicates evaluate to FALSE). The confidences returned for each class from each sibling Node
* that was scored are weighted by the proportion of the number of records in that Node, then summed to produce a
* total confidence for each class. The winner is the class with the highest confidence. Note that
* weightedConfidence should be applied recursively to deal with situations where several predicates within the
* tree evaluate to UNKNOWN during the scoring of a case.
* - aggregateNodes: If a Node's predicate value evaluates to UNKNOWN while traversing the tree, we consider
* evaluation of the Node's predicate being TRUE and follow this Node. In addition, subsequent Nodes to the initial
* Node are evaluated as well. This procedure is applied recursively for each Node being evaluated until a leaf
* Node is reached. All leaf Nodes being reached by this procedure are aggregated such that for each value
* attribute of such a leaf Node's ScoreDistribution element the corresponding recordCount attribute values are
* accumulated. The value associated with the highest recordCount accumulated through this procedure is predicted.
* The basic idea of missingValueStrategy aggregateNodes is to aggregate all leaf Nodes which may be reached by a
* record with one or more missing values considering all possible values. Strategy aggregateNodes calculates a
* virtual Node and predicts a score according to this virtual Node. Requires the presence of attribute recordCount
* in all ScoreDistribution elements.
* - none: Comparisons with missing values other than checks for missing values always evaluate to FALSE. If no
* rule fires, then use the noTrueChildStrategy to decide on a result. This option requires that missing values be
* handled after all rules at the Node have been evaluated.
* Note: In contrast to lastPrediction, evaluation is carried on instead of stopping immediately upon first
* discovery of a Node who's predicate value cannot be determined due to missing values.
*/
val lastPrediction, nullPrediction, defaultChild, weightedConfidence, aggregateNodes, none = Value
}
/**
* Defines what to do in situations where scoring cannot reach a leaf node.
*/
object NoTrueChildStrategy extends Enumeration {
type NoTrueChildStrategy = Value
val returnNullPrediction, returnLastPrediction = Value
}
/**
* Indicates whether non-leaf Nodes in the tree model have exactly two children, or an unrestricted number of children.
*/
object SplitCharacteristic extends Enumeration {
type SplitCharacteristic = Value
val binarySplit, multiSplit = Value
}
import org.pmml4s.model.MissingValueStrategy._
import org.pmml4s.model.NoTrueChildStrategy._
import org.pmml4s.model.SplitCharacteristic._
trait HasTreeAttributes extends HasModelAttributes {
/**
* Defines a strategy for dealing with missing values.
*/
def missingValueStrategy: MissingValueStrategy
/**
* Defines a penalty applied to confidence calculation when missing value handling is performed.
*/
def missingValuePenalty: Double
/**
* Defines what to do in situations where scoring cannot reach a leaf node.
*/
def noTrueChildStrategy: NoTrueChildStrategy
/**
* Indicates whether non-leaf Nodes in the tree model have exactly two children, or an unrestricted number of
* children.
* In the case of multiSplit, it means that each Node may have 0 or more child Nodes. In the case of binarySplit,
* it means that each Node must have either 0 or 2 child Nodes.
*/
def splitCharacteristic: SplitCharacteristic
}
/**
* Holds attributes of a Tree model
*/
class TreeAttributes(
override val functionName: MiningFunction,
override val modelName: Option[String] = None,
override val algorithmName: Option[String] = None,
override val isScorable: Boolean = true,
val missingValueStrategy: MissingValueStrategy = MissingValueStrategy.none,
val missingValuePenalty: Double = 1.0,
val noTrueChildStrategy: NoTrueChildStrategy = NoTrueChildStrategy.returnNullPrediction,
val splitCharacteristic: SplitCharacteristic = SplitCharacteristic.multiSplit)
extends ModelAttributes(functionName, modelName, algorithmName, isScorable) with HasTreeAttributes
trait HasWrappedTreeAttributes extends HasWrappedModelAttributes with HasTreeAttributes {
override def attributes: TreeAttributes
def missingValueStrategy: MissingValueStrategy = attributes.missingValueStrategy
def missingValuePenalty: Double = attributes.missingValuePenalty
def noTrueChildStrategy: NoTrueChildStrategy = attributes.noTrueChildStrategy
def splitCharacteristic: SplitCharacteristic = attributes.splitCharacteristic
}
/**
* This element is an encapsulation for either defining a split or a leaf in a tree model.
* Every Node contains a predicate that identifies a rule for choosing itself or any of its siblings.
* A predicate may be an expression composed of other nested predicates.
*/
class Node(
val predicate: Predicate,
val children: Array[Node],
val id: Option[String] = None,
val score: Option[Any] = None,
val recordCount: Option[Double] = None,
val defaultChild: Option[String] = None,
val scoreDistributions: ScoreDistributions = new ScoreDistributions,
val partition: Option[Partition] = None,
val embeddedModel: Option[EmbeddedModel] = None) extends Predicate with HasScoreDistributions {
def eval(series: Series): Predication = predicate.eval(series)
val isLeaf: Boolean = children.isEmpty
val isSplit: Boolean = children.nonEmpty
val size: Int = children.length
def apply(i: Int): Node = children(i)
val defaultChildNode: Option[Node] = defaultChild.flatMap {
x =>
children.find {
y => y.id match {
case Some(id) => id == x
case None => false
}
}
}
}
class TreeOutputs extends MixedClsWithRegOutputs with MutableConfidence with MutableEntityId
