scala-sparkML学习笔记：struct type tinyint size int indices array int values array double type

错误类型：

CSV data source does not support struct<type:tinyint,size:int,indices:array<int>,values:array<double>> data type.

predictPredict.select("user_id", "probability", "label").coalesce(1) 
          .write.format("com.databricks.spark.csv").mode("overwrite") 
          .option("header", "true").option("delimiter","\t").option("nullValue", Const.NULL) 
          .save(fileName.predictResultFile + day) 

predictPredict选择probability列保存会出现'`probability`' is of struct<type:tinyint,size:int,indices:array<int>,values:array<double>> type 这个错误， 因为是DenseVector不可以直接报保存到csv文件， 可以有下面两种解决方法： （主要思想是选择DenseVector中预测为1的那一列，类型为double）

        /*
        import org.apache.spark.sql.SparkSession
        val spark = SparkSession.builder().config("spark.debug.maxToStringFields", 500).enableHiveSupport.appName("QDSpark Pipeline").getOrCreate()
        import spark.implicits._

        val probabilityDataFrame = predictPredict.select("user_id", "probability", "label").rdd.map( row => (row.getInt(0), row.getAs[DenseVector](1)(1), row.getDouble(2)) ).toDF

        probabilityDataFrame.select("_1", "_2", "_3").coalesce(1) 
          .write.format("com.databricks.spark.csv").mode("overwrite") 
          .option("header", "true").option("delimiter","\t").option("nullValue", Const.NULL) 
          .save(fileName.predictResultFile + day) 
        */
        
        val stages = new ArrayBuffer[StructField]() 
        stages += StructField("user_id", IntegerType, true) 
        stages += StructField("probability", DoubleType, true) 
        stages += StructField("label", DoubleType, true) 
        val schema = new StructType( stages.toArray  )
        val probabilityNewRDD = predictPredict.select("user_id", "probability", "label").rdd.map( row => Row(row.getInt(0), row.getAs[DenseVector](1)(1), row.getDouble(2)) )
        val probabilityDataFrame = SparkConfTrait.spark.createDataFrame(probabilityNewRDD, schema)

        probabilityDataFrame.select("user_id", "probability", "label").coalesce(1) 
          .write.format("com.databricks.spark.csv").mode("overwrite") 
          .option("header", "true").option("delimiter","\t").option("nullValue", Const.NULL) 
          .save(fileName.predictResultFile + day)