Применяет MAP () к Dataframe, всегда преобразует DF в RDD.
scala> val custDF = sqlContext.read.format("com.databricks.spark.avro").load("/user/cloudera/practice1/problem7/customer/avro")
**custDF: org.apache.spark.sql.DataFrame** = [customer_id: int, customer_fname: string, customer_lname: string]
scala> val a = custDF.map(x=>x(0)+"\t"+x(1)+"\t"+x(2))
**a: org.apache.spark.rdd.RDD[String]** = MapPartitionsRDD[106] at map at <console>:36