take(N):随机取RDD中N个元素
scala> s3.take(4)
res4: Array[Int] = Array(1, 2, 3, 4)
takeOrder(N):升序取出RDD中N个元素
scala> s3.takeOrdered(2)
res7: Array[Int] = Array(1, 2)
top(N):降序取出RDD中N个元素
scala> s3.top(4)
res8: Array[Int] = Array(7, 6, 5, 4)
map()
返回：一个新的RDD
scala> val rrr = sc.parallelize(List(1,2,3,3))
scala> val rrr=rdd.map(x => x +1 )
scala> rrr.collect()
Array[Int] = Array(2, 3, 4, 4)
scala> rrr.foreach(println)//打印，结果似乎是无序的
scala> val rdd2=sc.makeRDD(List(1,2,3,4))
scala> rdd2.map(x=>1 to x).collect
res22: Array[scala.collection.immutable.Range.Inclusive] = Array(Range(1), Range(1, 2), Range(1, 2, 3), Range(1, 2, 3, 4))
flatMap()：对RDD中的每一个元素经过指定函数运算产生若干个新元素，所有新元素构成新的RDD
scala> rdd2.flatMap(x=>1 to x).collect
res21: Array[Int] = Array(1, 1, 2, 1, 2, 3, 1, 2, 3, 4)
flatMap()与map()的区别：
scala> t1.collect
res18: Array[String] = Array(hello hadoop world, hello spark hive sql nihao vim, linux nihao spark xiaomi, world xiaomi hive vim test, "this is spark test ", "hello hadoop is could ", "OpenStack is could could ")
//每个逗号分割的是一行
scala> val t2 = t1.map(x=>x.split(" "))
t2: org.apache.spark.rdd.RDD[Array[String]] = MapPartitionsRDD[17] at map at <console>:25
scala> t2.collect
res19: Array[Array[String]] = Array(Array(hello, hadoop, world), Array(hello, spark, hive, sql, nihao, vim), Array(linux, nihao, spark, xiaomi), Array(world, xiaomi, hive, vim, test), Array(this, is, spark, test), Array(hello, hadoop, is, could), Array(OpenStack, is, could, could))
//每一行里又建立了一个Array
scala> val t3=t1.flatMap(x=>x.split(" "))
t3: org.apache.spark.rdd.RDD[String] = MapPartitionsRDD[18] at flatMap at <console>:25
scala> t3.collect
res20: Array[String] = Array(hello, hadoop, world, hello, spark, hive, sql, nihao, vim, linux, nihao, spark, xiaomi, world, xiaomi, hive, vim, test, this, is, spark, test, hello, hadoop, is, could, OpenStack, is, could, could)
//flatMap做的是扁平化
filter()
对每个元素进行筛选，返回符合条件的元素组成的一个新RDD
scala> val rdd = sc.parallelize(List(1,2,3,3))
scala> rdd.filter(x => x != 1).collect()
Array[Int] = Array(2, 3, 3)
countApprox(超时时间) :统计近似元素个数
sortBy()排序
scala> val d1=sc.makeRDD(List(3,1,4,2,6,9))
scala> d1.sortBy(x=>x) //默认升序
scala> res30.collect
res31: Array[Int] = Array(1, 2, 3, 4, 6, 9)
scala> d1.sortBy(x=>x,false)
scala> res32.collect
res33: Array[Int] = Array(9, 6, 4, 3, 2, 1)
scala> val l1=sc.makeRDD(List(("cat",21),("dog",2),("pig",3)))
scala> l1.sortBy(x=>x._2).collect
res69: Array[(String, Int)] = Array((dog,2), (pig,3), (cat,21))
scala> l1.sortBy(x=>x._1).collect
res70: Array[(String, Int)] = Array((cat,21), (dog,2), (pig,3)
countByValue() ： 各个元素出现的次数，返回Map类型
scala> val c=sc.makeRDD(List(1,3,1,4,1,3))
scala> c.countByValue
scala> s3.take(4)
res4: Array[Int] = Array(1, 2, 3, 4)
takeOrder(N):升序取出RDD中N个元素
scala> s3.takeOrdered(2)
res7: Array[Int] = Array(1, 2)
top(N):降序取出RDD中N个元素
scala> s3.top(4)
res8: Array[Int] = Array(7, 6, 5, 4)
map()
返回：一个新的RDD
scala> val rrr = sc.parallelize(List(1,2,3,3))
scala> val rrr=rdd.map(x => x +1 )
scala> rrr.collect()
Array[Int] = Array(2, 3, 4, 4)
scala> rrr.foreach(println)//打印，结果似乎是无序的
scala> val rdd2=sc.makeRDD(List(1,2,3,4))
scala> rdd2.map(x=>1 to x).collect
res22: Array[scala.collection.immutable.Range.Inclusive] = Array(Range(1), Range(1, 2), Range(1, 2, 3), Range(1, 2, 3, 4))
flatMap()：对RDD中的每一个元素经过指定函数运算产生若干个新元素，所有新元素构成新的RDD
scala> rdd2.flatMap(x=>1 to x).collect
res21: Array[Int] = Array(1, 1, 2, 1, 2, 3, 1, 2, 3, 4)
flatMap()与map()的区别：
scala> t1.collect
res18: Array[String] = Array(hello hadoop world, hello spark hive sql nihao vim, linux nihao spark xiaomi, world xiaomi hive vim test, "this is spark test ", "hello hadoop is could ", "OpenStack is could could ")
//每个逗号分割的是一行
scala> val t2 = t1.map(x=>x.split(" "))
t2: org.apache.spark.rdd.RDD[Array[String]] = MapPartitionsRDD[17] at map at <console>:25
scala> t2.collect
res19: Array[Array[String]] = Array(Array(hello, hadoop, world), Array(hello, spark, hive, sql, nihao, vim), Array(linux, nihao, spark, xiaomi), Array(world, xiaomi, hive, vim, test), Array(this, is, spark, test), Array(hello, hadoop, is, could), Array(OpenStack, is, could, could))
//每一行里又建立了一个Array
scala> val t3=t1.flatMap(x=>x.split(" "))
t3: org.apache.spark.rdd.RDD[String] = MapPartitionsRDD[18] at flatMap at <console>:25
scala> t3.collect
res20: Array[String] = Array(hello, hadoop, world, hello, spark, hive, sql, nihao, vim, linux, nihao, spark, xiaomi, world, xiaomi, hive, vim, test, this, is, spark, test, hello, hadoop, is, could, OpenStack, is, could, could)
//flatMap做的是扁平化
filter()
对每个元素进行筛选，返回符合条件的元素组成的一个新RDD
scala> val rdd = sc.parallelize(List(1,2,3,3))
scala> rdd.filter(x => x != 1).collect()
Array[Int] = Array(2, 3, 3)
countApprox(超时时间) :统计近似元素个数
sortBy()排序
scala> val d1=sc.makeRDD(List(3,1,4,2,6,9))
scala> d1.sortBy(x=>x) //默认升序
scala> res30.collect
res31: Array[Int] = Array(1, 2, 3, 4, 6, 9)
scala> d1.sortBy(x=>x,false)
scala> res32.collect
res33: Array[Int] = Array(9, 6, 4, 3, 2, 1)
scala> val l1=sc.makeRDD(List(("cat",21),("dog",2),("pig",3)))
scala> l1.sortBy(x=>x._2).collect
res69: Array[(String, Int)] = Array((dog,2), (pig,3), (cat,21))
scala> l1.sortBy(x=>x._1).collect
res70: Array[(String, Int)] = Array((cat,21), (dog,2), (pig,3)
countByValue() ： 各个元素出现的次数，返回Map类型
scala> val c=sc.makeRDD(List(1,3,1,4,1,3))
scala> c.countByValue
res75: scala.collection.Map[Int,Long] = Map(4 -> 1, 1 -> 3, 3 -> 2)
欢迎阅读者加qq1204738320交流学习