スパーク決定木分類器

    import org.apache.spark.mllib.tree.DecisionTree
import org.apache.spark.mllib.tree.model.DecisionTreeModel
import org.apache.spark.mllib.util.MLUtils
import org.apache.spark.mllib.linalg.Vectors
import org.apache.spark.mllib.regression.LabeledPoint
import org.apache.spark.mllib.feature.Normalizer
// read the data file and split it by spaces
val data = sc.textFile("/home/walker/mileage.txt").map( l => l.split("\\s+"))
// convert every string to doubles
var dbl = data.map (m => m.map(l => l.toDouble))
// Here we create a Dense Vector and LabeledPoint. A Dense Vector is a
// special kind of Vector than the scala regular Vector. So use Vectors, with an “s”
// at the end. The values must be doubles. A Dense Vector cannot handle
// blank values. If we had that those, we would use a Sparse Vector.
//
// The LabeledPoint is an object with the label and features. The label
// is 0, 1, 2 or gas mileage and the features are the weight, year, etc.
val parsedData = dbl.map { y =>
val c = Vectors.dense(y(1),y(2),y(3),y(4),y(5),y(6),y(7))
LabeledPoint(y(0), c)
}.cache()
// numClasses means number of classifications. In this case it is
// 0, 1, or 2
val numClasses = 3
// Here we say that the 6th feature is categorical with 3 possible values.
// It can be American, European, or Asian.
// The rest of the values are continuous, meaning just numbers and not discrete
// values.
val categoricalFeaturesInfo = Map(6 -> 3)
val impurity = "gini"
val maxDepth = 9
val maxBins = 7
// Now feed the data into the model.
val model = DecisionTree.trainClassifier(parsedData, numClasses, categoricalFeaturesInfo , impurity, maxDepth, maxBins)
// Print out the results
println("Learned classification tree model:\n" + model.toDebugString)

    If (feature 3 <= 2930.0)
If (feature 1 <= 91.0)
Predict: 2.0
Else (feature 1 > 91.0)
If (feature 5 <= 81.0)
If (feature 2 <= 105.0)
Predict: 1.0
Else (feature 2 > 105.0)
If (feature 0 <= 4.0)
Predict: 0.0
Else (feature 0 > 4.0)
Predict: 1.0
Else (feature 5 > 81.0)
If (feature 2 <= 85.0)
Predict: 0.0
Else (feature 2 > 85.0)
Predict: 2.0
Else (feature 3 > 2930.0)
If (feature 3 <= 3193.0)
If (feature 2 <= 85.0)
Predict: 0.0
Else (feature 2 > 85.0)
If (feature 2 <= 105.0)
If (feature 0 <= 4.0)
Predict: 1.0
Else (feature 0 > 4.0)
Predict: 0.0
Else (feature 2 > 105.0)
Predict: 0.0
Else (feature 3 > 3193.0)
Predict: 0.0