Spark算子总结及案例
spark算子大致上可分三大类算子：
1、Value数据类型的Transformation算子，这种变换不触发提交作业，针对处理的数据项是Value型的数据。
2、Key-Value数据类型的Transformation算子，这种变换不触发提交作业，针对处理的数据项是Key-Value型的数据。
3、Action算子，这类算子会触发SparkContext提交作业。
一、Value型Transformation算子
1）map
val a = sc.parallelize(List("dog", "salmon", "salmon", "rat", "elephant"), )
val b = a.map(_.length)
val c = a.zip(b)
c.collect
res0: Array[(String, Int)] = Array((dog,), (salmon,), (salmon,), (rat,), (elephant,))
2）flatMap
val a = sc.parallelize( to , )
a.flatMap( to _).collect
res47: Array[Int] = Array(, , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , ) sc.parallelize(List(, , ), ).flatMap(x => List(x, x, x)).collect
res85: Array[Int] = Array(, , , , , , , , )
3）mapPartiions
val x = sc.parallelize( to , )
x.flatMap(List.fill(scala.util.Random.nextInt())(_)).collect res1: Array[Int] = Array(, , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , )
4）glom（形成一个Array数组）
val a = sc.parallelize( to , )
a.glom.collect
res8: Array[Array[Int]] = Array(Array(, , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , ), Array(, , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , ), Array(, , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , ))
5）union
val a = sc.parallelize( to , )
val b = sc.parallelize( to , )
(a ++ b).collect
res0: Array[Int] = Array(, , , , , )
6）cartesian（笛卡尔操作）
val x = sc.parallelize(List(,,,,))
val y = sc.parallelize(List(,,,,))
x.cartesian(y).collect
res0: Array[(Int, Int)] = Array((,), (,), (,), (,), (,), (,), (,), (,), (,), (,), (,), (,), (,), (,), (,), (,), (,), (,), (,), (,), (,), (,), (,), (,), (,))
7）groupBy（生成相应的key，相同的放在一起）
val a = sc.parallelize( to , )
a.groupBy(x => { if (x % == ) "even" else "odd" }).collect
res42: Array[(String, Seq[Int])] = Array((even,ArrayBuffer(, , , )), (odd,ArrayBuffer(, , , , )))
8）filter
val a = sc.parallelize( to , )
val b = a.filter(_ % == )
b.collect
res3: Array[Int] = Array(, , , , )
9）distinct（去重）
val c = sc.parallelize(List("Gnu", "Cat", "Rat", "Dog", "Gnu", "Rat"), )
c.distinct.collect
res6: Array[String] = Array(Dog, Gnu, Cat, Rat)
10）subtract（去掉含有重复的项）
val a = sc.parallelize( to , )
val b = sc.parallelize( to , )
val c = a.subtract(b)
c.collect
res3: Array[Int] = Array(, , , , , )
11）sample
val a = sc.parallelize( to , )
a.sample(false, 0.1, ).count
res24: Long =
12）takesample
val x = sc.parallelize( to , )
x.takeSample(true, , )
res3: Array[Int] = Array(, , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , )
13）cache、persist
val c = sc.parallelize(List("Gnu", "Cat", "Rat", "Dog", "Gnu", "Rat"), )
c.getStorageLevel
res0: org.apache.spark.storage.StorageLevel = StorageLevel(false, false, false, false, )
c.cache
c.getStorageLevel
res2: org.apache.spark.storage.StorageLevel = StorageLevel(false, true, false, true, )
二、Key-Value型Transformation算子
1）mapValues
val a = sc.parallelize(List("dog", "tiger", "lion", "cat", "panther", "eagle"), )
val b = a.map(x => (x.length, x))
b.mapValues("x" + _ + "x").collect
res5: Array[(Int, String)] = Array((,xdogx), (,xtigerx), (,xlionx), (,xcatx), (,xpantherx), (,xeaglex))
2）combineByKey
val a = sc.parallelize(List("dog","cat","gnu","salmon","rabbit","turkey","wolf","bear","bee"), )
val b = sc.parallelize(List(,,,,,,,,), )
val c = b.zip(a)
val d = c.combineByKey(List(_), (x:List[String], y:String) => y :: x, (x:List[String], y:List[String]) => x ::: y)
d.collect
res16: Array[(Int, List[String])] = Array((,List(cat, dog, turkey)), (,List(gnu, rabbit, salmon, bee, bear, wolf)))
3）reduceByKey
val a = sc.parallelize(List("dog", "cat", "owl", "gnu", "ant"), )
val b = a.map(x => (x.length, x))
b.reduceByKey(_ + _).collect
res86: Array[(Int, String)] = Array((,dogcatowlgnuant))
val a = sc.parallelize(List("dog", "tiger", "lion", "cat", "panther", "eagle"), )
val b = a.map(x => (x.length, x))
b.reduceByKey(_ + _).collect
res87: Array[(Int, String)] = Array((,lion), (,dogcat), (,panther), (,tigereagle))
4）partitionBy
（对RDD进行分区操作）
5）cogroup
val a = sc.parallelize(List(, , , ), )
val b = a.map((_, "b"))
val c = a.map((_, "c"))
b.cogroup(c).collect
res7: Array[(Int, (Iterable[String], Iterable[String]))] = Array(
(,(ArrayBuffer(b),ArrayBuffer(c))),
(,(ArrayBuffer(b),ArrayBuffer(c))),
(,(ArrayBuffer(b, b),ArrayBuffer(c, c)))
)
6）join
val a = sc.parallelize(List("dog", "salmon", "salmon", "rat", "elephant"), )
val b = a.keyBy(_.length)
val c = sc.parallelize(List("dog","cat","gnu","salmon","rabbit","turkey","wolf","bear","bee"), )
val d = c.keyBy(_.length)
b.join(d).collect
res0: Array[(Int, (String, String))] = Array((,(salmon,salmon)), (,(salmon,rabbit)), (,(salmon,turkey)), (,(salmon,salmon)), (,(salmon,rabbit)), (,(salmon,turkey)), (,(dog,dog)), (,(dog,cat)), (,(dog,gnu)), (,(dog,bee)), (,(rat,dog)), (,(rat,cat)), (,(rat,gnu)), (,(rat,bee)))
7）leftOutJoin
val a = sc.parallelize(List("dog", "salmon", "salmon", "rat", "elephant"), )
val b = a.keyBy(_.length)
val c = sc.parallelize(List("dog","cat","gnu","salmon","rabbit","turkey","wolf","bear","bee"), )
val d = c.keyBy(_.length)
b.leftOuterJoin(d).collect
res1: Array[(Int, (String, Option[String]))] = Array((,(salmon,Some(salmon))), (,(salmon,Some(rabbit))), (,(salmon,Some(turkey))), (,(salmon,Some(salmon))), (,(salmon,Some(rabbit))), (,(salmon,Some(turkey))), (,(dog,Some(dog))), (,(dog,Some(cat))), (,(dog,Some(gnu))), (,(dog,Some(bee))), (,(rat,Some(dog))), (,(rat,Some(cat))), (,(rat,Some(gnu))), (,(rat,Some(bee))), (,(elephant,None)))
8）rightOutJoin
val a = sc.parallelize(List("dog", "salmon", "salmon", "rat", "elephant"), )
val b = a.keyBy(_.length)
val c = sc.parallelize(List("dog","cat","gnu","salmon","rabbit","turkey","wolf","bear","bee"), )
val d = c.keyBy(_.length)
b.rightOuterJoin(d).collect
res2: Array[(Int, (Option[String], String))] = Array((,(Some(salmon),salmon)), (,(Some(salmon),rabbit)), (,(Some(salmon),turkey)), (,(Some(salmon),salmon)), (,(Some(salmon),rabbit)), (,(Some(salmon),turkey)), (,(Some(dog),dog)), (,(Some(dog),cat)), (,(Some(dog),gnu)), (,(Some(dog),bee)), (,(Some(rat),dog)), (,(Some(rat),cat)), (,(Some(rat),gnu)), (,(Some(rat),bee)), (,(None,wolf)), (,(None,bear)))
三、Actions算子
1）foreach
val c = sc.parallelize(List("cat", "dog", "tiger", "lion", "gnu", "crocodile", "ant", "whale", "dolphin", "spider"), )
c.foreach(x => println(x + "s are yummy"))
lions are yummy
gnus are yummy
crocodiles are yummy
ants are yummy
whales are yummy
dolphins are yummy
spiders are yummy
2）saveAsTextFile
val a = sc.parallelize( to , )
a.saveAsTextFile("mydata_a")
// :: INFO FileOutputCommitter: Saved output of task 'attempt_201404032111_0000_m_000002_71' to file:/home/cloudera/Documents/spark-0.9.-incubating-bin-cdh4/bin/mydata_a
3）saveAsObjectFile
val x = sc.parallelize( to , )
x.saveAsObjectFile("objFile")
val y = sc.objectFile[Int]("objFile")
y.collect
res52: Array[Int] = Array[Int] = Array(, , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , )
4）collect
val c = sc.parallelize(List("Gnu", "Cat", "Rat", "Dog", "Gnu", "Rat"), )
c.collect
res29: Array[String] = Array(Gnu, Cat, Rat, Dog, Gnu, Rat)
5）collectAsMap
val a = sc.parallelize(List(, , , ), )
val b = a.zip(a)
b.collectAsMap
res1: scala.collection.Map[Int,Int] = Map( -> , -> , -> )
6）reduceByKeyLocally
val a = sc.parallelize(List("dog", "cat", "owl", "gnu", "ant"), )
val b = a.map(x => (x.length, x))
b.reduceByKey(_ + _).collect
res86: Array[(Int, String)] = Array((,dogcatowlgnuant))
7）lookup
val a = sc.parallelize(List("dog", "tiger", "lion", "cat", "panther", "eagle"), )
val b = a.map(x => (x.length, x))
b.lookup()
res0: Seq[String] = WrappedArray(tiger, eagle)
8）count
val c = sc.parallelize(List("Gnu", "Cat", "Rat", "Dog"), )
c.count
res2: Long =
9）top
val c = sc.parallelize(Array(, , , , , ), )
c.top()
res28: Array[Int] = Array(, )
10）reduce
val a = sc.parallelize( to , )
a.reduce(_ + _)
res41: Int =
11）fold
val a = sc.parallelize(List(,,), )
a.fold()(_ + _)
res59: Int =
12）aggregate
val z = sc.parallelize(List(,,,,,), ) // lets first print out the contents of the RDD with partition labels
def myfunc(index: Int, iter: Iterator[(Int)]) : Iterator[String] = {
iter.toList.map(x => "[partID:" + index + ", val: " + x + "]").iterator
} z.mapPartitionsWithIndex(myfunc).collect
res28: Array[String] = Array([partID:, val: ], [partID:, val: ], [partID:, val: ], [partID:, val: ], [partID:, val: ], [partID:, val: ]) z.aggregate()(math.max(_, _), _ + _)
res40: Int =
参考：http://homepage.cs.latrobe.edu.au/zhe/ZhenHeSparkRDDAPIExamples.html
Spark算子总结及案例的更多相关文章
- Spark算子总结（带案例）
Spark算子总结(带案例) spark算子大致上可分三大类算子: 1.Value数据类型的Transformation算子,这种变换不触发提交作业,针对处理的数据项是Value型的数据. 2.Key ...
- Spark Streaming 进阶与案例实战
Spark Streaming 进阶与案例实战 1.带状态的算子: UpdateStateByKey 2.实战:计算到目前位置累积出现的单词个数写入到MySql中 1.create table CRE ...
- (转)Spark 算子系列文章
http://lxw1234.com/archives/2015/07/363.htm Spark算子:RDD基本转换操作(1)–map.flagMap.distinct Spark算子:RDD创建操 ...
- 《图解Spark：核心技术与案例实战》作者经验谈
1,看您有维护博客,还利用业余时间著书,在技术输出.自我提升以及本职工作的时间利用上您有没有什么心得和大家分享?(也可以包含一些您写书的小故事.)回答:在工作之余能够写博客.著书主要对技术的坚持和热爱 ...
- UserView--第二种方式（避免第一种方式Set饱和），基于Spark算子的java代码实现
UserView--第二种方式(避免第一种方式Set饱和),基于Spark算子的java代码实现 测试数据 java代码 package com.hzf.spark.study; import ...
- UserView--第一种方式set去重，基于Spark算子的java代码实现
UserView--第一种方式set去重,基于Spark算子的java代码实现 测试数据 java代码 package com.hzf.spark.study; import java.util.Ha ...
- 【原创 Hadoop&Spark 动手实践 6】Spark 编程实例与案例演示
[原创 Hadoop&Spark 动手实践 6]Spark 编程实例与案例演示 Spark 编程实例和简易电影分析系统的编写 目标: 1. 掌握理论:了解Spark编程的理论基础 2. 搭建 ...
- Scala进阶之路-Spark底层通信小案例
Scala进阶之路-Spark底层通信小案例 作者:尹正杰 版权声明:原创作品,谢绝转载!否则将追究法律责任. 一.Spark Master和worker通信过程简介 1>.Worker会向ma ...
- spark算子之DataFrame和DataSet
前言 传统的RDD相对于mapreduce和storm提供了丰富强大的算子.在spark慢慢步入DataFrame到DataSet的今天,在算子的类型基本不变的情况下,这两个数据集提供了更为强大的的功 ...
随机推荐
- 一个combineInputformat
mark import java.io.DataInput; import java.io.DataOutput; import java.io.IOException; import org.apa ...
- MIME小知识
http://www.alixixi.com/program/a/2008020514775.shtml 用户可以通过使用MIME以设置服务器传送多媒体如声音和动画信息,这一切可能通过CGI脚本来进行 ...
- 畅通工程续 (dijkstra)
畅通工程续 Time Limit: 3000/1000 MS (Java/Others) Memory Limit: 32768/32768 K (Java/Others)Total Submi ...
- 并发数据（锁）ReaderWriterLockSlim
ReaderWriterLockSlim 类 ReaderWriterLockSlim 类支持三种锁定模式:Read,Write,UpgradeableRead.这三种模式对应的方法分别是 Enter ...
- how to enable #ifdef macro in the command line of make?
Compilers normally use the -D flags eg Code: test.o: test.cpp $(CC) $(CFLAGS) -DTEST1 test.cpp -o $@
- CSS——z-index
1.特性 ①z-index的值表示谁压着谁.值大的压住值小的. ②只有定位的元素才有z-index 值,也就是说,浮动不可以,相对定位,绝对定位,固定定位都可以. ③z-index值没有单位,是一个正 ...
- LVS详解
v\:* {behavior:url(#default#VML);} o\:* {behavior:url(#default#VML);} w\:* {behavior:url(#default#VM ...
- Spring MVC Json数据传递
json是一种常见的传递格式,是一种键值对应的格式.并且数据大小会比较小,方便传递.所以在开发中经常会用到json. 首先看一下json的格式: {key1:value1,key2:value2} 每 ...
- Analyzing the Meaning of Sentences
1. How can we represent natural language meaning so that a computer can process these representation ...
- mysql--学生课程成绩表
创建表student: CREATE TABLE `student` ( `sid` INT(11) NOT NULL AUTO_INCREMENT, `sname` VARCHAR(20) NOT ...