spark-mlib線性迴歸

import org.apache.spark.SparkConf
import org.apache.spark.SparkContext
import org.apache.spark.ml.feature.VectorAssembler
import org.apache.spark.sql.SQLContext
import org.apache.spark.ml.regression.LinearRegression
object SparkMlib {
  def main(args: Array[String]): Unit = {
    val conf = new SparkConf().setAppName("mlib").setMaster("local")
    
    val context = new SparkContext(conf)
    val sqlContext = new SQLContext(context)
    
    val rdd = context.makeRDD(List((1,3,9),(2,6,18),(3,9,27),(4,12,36)))
    val cols = Array("x1","x2")
    val vectors = new VectorAssembler().setInputCols(cols).setOutputCol("predict")
    
    import sqlContext.implicits._
    val x = vectors.transform(rdd.toDF("x1","x2","y"))
    
    val model = new LinearRegression()
                    //自變數的資料名
                    .setFeaturesCol("predict")
                    //因變數
                    .setLabelCol("y")
                    //是否有截距
                    .setFitIntercept(false)
                    //訓練模型
                    .fit(x)
                    
     //線性迴歸的係數              
     println(model.coefficients)
     //線性迴歸的截距
     println(model.intercept)
     //線性迴歸的自變數的個數
     println(model.numFeatures)
     //上面的feature列
     println(model.summary.featuresCol)
     //r2
     println(model.summary.r2)
     //平均絕對誤差
     println(model.summary.meanAbsoluteError)
     //方差
     println(model.summary.meanSquaredError)
     //新的集合x1,x2 預測y
     val testRdd  = context.makeRDD(List((1,3),(2,6),(3,9),(4,12)))
     //根據上面的模型預測結果
     val testSet = vectors.transform(testRdd.toDF("x1","x2"))
     val pre = model.transform(testSet)
    
     pre.show()
     //println(pre.predictions)
    
  }
}