spark 如何获得分类概率

在进行分类时，通常不仅希望知道该样本是被预测为0，1，还希望获得该样本被预测为0，1的概率 

LR中：

val model = new LogisticRegressionWithLBFGS().setNumClasses(2).run(trainingData)

model.clearThreshold()

//默认Threshold为0.5，只需通过model.clearThreshold()函数去掉阈值即可获得分类概率

GBDT 中：

原始的 predict函数只能输出0,1；我们需要通过以下原代码修改该函数即可得到概率

def predict(features: Vector): Double = {
  (algo, combiningStrategy) match {
    case (Regression, Sum) =>
      predictBySumming(features)
    case (Regression, Average) =>
      predictBySumming(features) / sumWeights
 case (Classification, Sum) => // binary classification
 val prediction = predictBySumming(features)
      // TODO: predicted labels are +1 or -1 for GBT. Need a better way to store this info.
 //if (prediction > 0.0) 1.0 else 0.0（原始）
 (1/(1+math.pow(2.7, -prediction)))//修改为sigmoid函数
    case (Classification, Vote) =>
      predictByVoting(features)
    case _ =>
      throw new IllegalArgumentException(
        "TreeEnsembleModel1 given unsupported (algo, combiningStrategy) combination: " +
          s"($algo, $combiningStrategy).")
  }
}

然后import重定义的class即可