Spark ML 文本的分类

最近一直在研究Spark的分类算法，因为我们是做日志文本分类，在官网和各大网站一直没找到相应的Demo，经过1个多月的研究，终于有点成效。

val sparkConf = new SparkConf().setAppName("DecisionTree1").setMaster("local[2]")

    val sc = new SparkContext(sparkConf)

    var data1 = sc.textFile("/XXX/sample_libsvm_data.txt")

    val hashingTF = new HashingTF()

    val data = data1.map { line =>
      val parts = line.split('\t')
      LabeledPoint(parts(0).toDouble, hashingTF.transform(parts.tail))
    }

    val splits = data.randomSplit(Array(0.9, 0.1))
    val (trainingData, testData) = (splits(0), splits(1))

    // Train a DecisionTree model.
    //  Empty categoricalFeaturesInfo indicates all features are continuous.
    val numClasses = 5
    val categoricalFeaturesInfo = Map[Int, Int]()
    val impurity = "gini"
    val maxDepth = 5
    val maxBins = 32

    println("--------------------train--------------------")

    val model = DecisionTree.trainClassifier(trainingData, numClasses, categoricalFeaturesInfo,
      impurity, maxDepth, maxBins)

    println("--------------------Test--------------------")

    // Evaluate model on test instances and compute test error
    val testStr = Array("l","o","k")

    val prediction = model.predict(hashingTF.transform(testStr))

    println("-----------------------------------------")
    println(prediction)
    println("-----------------------------------------")

样例数据：

2    f    g    k    m
3    o    p    s    d
4    i    l    o    v
4    i    l    o    w
4    i    l    o    f
4    i    l    o    k
4    i    l    o    n
4    i    l    o    a
2    f    g    i    m
2    f    g    o    m
2    f    g    u    m
2    f    g    w    m
3    o    k    s    d
3    o    m    s    d
3    o    s    s    d
3    o    i    s    d

Classification算法只支持Double类型，其实我们的核心就是怎么把字符串转成Double型的向量，在Spark1.3.0版本中有 HashingTF 来做转化，就发现程序很简单了。