XGBoost-4j by DMLC on Spark-1.6.1

Question

I am trying to use the XGBoost implementation by DMLC on Spark-1.6.1. I am able to train my data with XGBoost but facing difficulties in prediction. I actually want to do prediction in the way it can be done in Apache Spark mllib libraries, that helps with calculation of training error,precision, recall, specificity etc.

I am posting the codes below, also the error I am getting. I used this xgboost4j-spark-0.5-jar-with-dependencies.jar in spark-shell to start.

import org.apache.spark.mllib.regression.LabeledPoint
import org.apache.spark.mllib.linalg.Vectors
import org.apache.spark.mllib.util.MLUtils
import org.apache.spark.SparkContext._
import org.apache.spark.mllib.linalg.Vectors
import org.apache.spark.mllib.regression.LabeledPoint
import ml.dmlc.xgboost4j.scala.Booster
import ml.dmlc.xgboost4j.scala.spark.XGBoost
import ml.dmlc.xgboost4j.scala.DMatrix
import ml.dmlc.xgboost4j.scala.{Booster, DMatrix}
import ml.dmlc.xgboost4j.scala.spark.{DataUtils, XGBoost}
import org.apache.spark.{SparkConf, SparkContext}




//Load and parse the data file.
val data = sc.textFile("file:///home/partha/credit_approval_2_attr.csv")
val data1 = sc.textFile("file:///home/partha/credit_app_fea.csv")


val parsedData = data.map { line =>
    val parts = line.split(',').map(_.toDouble)
    LabeledPoint(parts(0), Vectors.dense(parts.tail))
}.cache()

val parsedData1 = data1.map { line =>
    val parts = line.split(',').map(_.toDouble)
    Vectors.dense(parts)
}



//Tuning Parameters
val paramMap = List(
      "eta" -> 0.1f,  
      "max_depth" -> 5,
      "num_class" -> 2,
      "objective" -> "multi:softmax" ,
      "colsample_bytree" -> 0.8,
       "alpha" -> 1,
       "subsample" -> 0.5).toMap

  //Training the model  
val numRound = 20
val model = XGBoost.train(parsedData, paramMap, numRound, nWorkers = 1)
val pred = model.predict(parsedData1)
pred.collect()

Output from pred:

res0: Array[Array[Array[Float]]] = Array(Array(Array(0.0), Array(1.0), Array(1.0), Array(1.0), Array(0.0), Array(0.0), Array(1.0), Array(1.0), Array(0.0), Array(1.0), Array(0.0), Array(0.0), Array(0.0), Array(1.0), Array(1.0), Array(1.0), Array(1.0), Array(1.0), Array(0.0), Array(1.0), Array(1.0), Array(0.0), Array(1.0), Array(1.0), Array(1.0), Array(1.0), Array(1.0), Array(1.0), Array(1.0), Array(1.0), Array(1.0), Array(1.0), Array(1.0), Array(1.0), Array(1.0), Array(1.0), Array(1.0), Array(1.0), Array(0.0), Array(1.0), Array(1.0), Array(1.0), Array(0.0), Array(1.0), Array(1.0), Array(1.0), Array(1.0), Array(1.0), Array(0.0), Array(0.0), Array(0.0), Array(0.0), Array(1.0), Array(0.0), Array(0.0), Array(0.0), Array(0.0), Array(0.0), Array(0.0), Array(1.0), Array(1.0), Array(1.0), Array(...

Now when I am using:

val labelAndPreds = parsedData.map { point =>
          val prediction = model.predict(point.features)
          (point.label, prediction)
        }

Output:

<console>:66: error: overloaded method value predict with alternatives:
  (testSet: ml.dmlc.xgboost4j.scala.DMatrix)Array[Array[Float]] <and>
  (testSet: org.apache.spark.rdd.RDD[org.apache.spark.mllib.linalg.Vector])org.apache.spark.rdd.RDD[Array[Array[Float]]]
 cannot be applied to (org.apache.spark.mllib.linalg.Vector)
                  val prediction = model.predict(point.features)
                                     ^

And then tried this, since predict requires an RDD[Vector].

val labelAndPreds1 = parsedData.map { point =>
          val prediction = model.predict(Vectors.dense(point.features))
          (point.label, prediction)
        }

The outcome was:

<console>:66: error: overloaded method value dense with alternatives:
  (values: Array[Double])org.apache.spark.mllib.linalg.Vector <and>
  (firstValue: Double,otherValues: Double*)org.apache.spark.mllib.linalg.Vector
 cannot be applied to (org.apache.spark.mllib.linalg.Vector)
                  val prediction = model.predict(Vectors.dense(point.features))
                                                         ^

Clearly its an issue of RDD type which I am trying to sort out, this is easy with GBT on spark (http://spark.apache.org/docs/latest/mllib-ensembles.html#gradient-boosted-trees-gbts).

Am I trying to do this the right way?

Any help or suggestion would be awesome.

dirceusemighini dirceusemighini · Accepted Answer · 2016-08-30T20:33:41

Actually this isn't available at XGboost algorithms. I'm facing the same problem here and have implemented the following method:

import ml.dmlc.xgboost4j.scala.spark.DataUtils // thanks to @Z Simon

def labelPredict(testSet: RDD[XGBLabeledPoint],
               useExternalCache: Boolean = false,
               booster: XGBoostModel): RDD[(Float, Float)] = {
val broadcastBooster = testSet.sparkContext.broadcast(booster)
testSet.mapPartitions { testData =>
  val (auxiliaryIterator, testDataIterator) = testData.duplicate
  val testDataArray = auxiliaryIterator.toArray
  val prediction = broadcastBooster.value.predict(new DMatrix(testDataIterator)).flatten
  testDataArray
    .zip(prediction)
    .map {
      case (labeledPoint, predictionValue) =>
        (labeledPoint.label, predictionValue)
    }.toIterator
}

}

This is almost the same code that XGBoost has actually, but it is using the label of labeledpoint in the predictions return. When you pass a Labeledpoint to this method, it will return a RDD of Tuple with (label, prediction) for each value.

XGBoost-4j by DMLC on Spark-1.6.1

2 Answers