upstream/piflow-bundle/src/main/scala/cn/piflow/bundle/ml_clustering/KmeansTraining.scala
1package cn.piflow.bundle.ml_clustering
2
3import cn.piflow.{JobContext, JobInputStream, JobOutputStream, ProcessContext}
4import cn.piflow.conf.{ConfigurableStop, Port, StopGroup}
5import cn.piflow.conf.bean.PropertyDescriptor
6import cn.piflow.conf.util.{ImageUtil, MapUtil}
7import org.apache.spark.ml.clustering.KMeans
8import org.apache.spark.sql.SparkSession
9
10class KmeansTraining extends ConfigurableStop{
11 val authorEmail: String = "06whuxx@163.com"
12 val description: String = "Kmeans clustering"
13 val inportList: List[String] = List(Port.DefaultPort)
14 val outportList: List[String] = List(Port.DefaultPort)
15 var training_data_path:String =_
16 var model_save_path:String=_
17 var maxIter:String=_
18 var k:Int=_
19 var minTol:String=_
20
21
22 def perform(in: JobInputStream, out: JobOutputStream, pec: JobContext): Unit = {
23 val spark = pec.get[SparkSession]()
24
25 //load data stored in libsvm format as a dataframe
26 val data=spark.read.format("libsvm").load(training_data_path)
27
28
29 //Param for maximum number of iterations (>= 0)
30 var maxIterValue:Int=50
31 if(maxIter!=""){
32 maxIterValue=maxIter.toInt
33 }
34
35 //Param for the convergence tolerance for iterative algorithms (>= 0).
36 var minTolValue:Double=1E-6
37 if(minTol!=""){
38 minTolValue=minTol.toDouble
39 }
40
41
42 //clustering with kmeans algorithm
43 val model=new KMeans()
44 .setMaxIter(maxIterValue)
45 .setTol(minTolValue)
46 .setK(k)
47 .fit(data)
48
49 //model persistence
50 model.save(model_save_path)
51
52 import spark.implicits._
53 val dfOut=Seq(model_save_path).toDF
54 dfOut.show()
55 out.write(dfOut)
56
57 }
58
59 def initialize(ctx: ProcessContext): Unit = {
60
61 }
62
63
64 def setProperties(map: Map[String, Any]): Unit = {
65 training_data_path=MapUtil.get(map,key="training_data_path").asInstanceOf[String]
66 model_save_path=MapUtil.get(map,key="model_save_path").asInstanceOf[String]
67 maxIter=MapUtil.get(map,key="maxIter").asInstanceOf[String]
68 minTol=MapUtil.get(map,key="minTol").asInstanceOf[String]
69 k=Integer.parseInt(MapUtil.get(map,key="k").toString)
70 }
71
72 override def getPropertyDescriptor(): List[PropertyDescriptor] = {
73 var descriptor : List[PropertyDescriptor] = List()
74 val training_data_path = new PropertyDescriptor().name("training_data_path").displayName("TRAINING_DATA_PATH").defaultValue("").required(true)
75 val model_save_path = new PropertyDescriptor().name("model_save_path").displayName("MODEL_SAVE_PATH").description("").defaultValue("").required(true)
76 val maxIter=new PropertyDescriptor().name("maxIter").displayName("MAX_ITER").description("Param for maximum number of iterations (>= 0).").defaultValue("").required(false)
77 val minTol=new PropertyDescriptor().name("minTol").displayName("MIN_TOL").description("Param for the convergence tolerance for iterative algorithms (>= 0).").defaultValue("").required(false)
78 val k=new PropertyDescriptor().name("k").displayName("K").description("The number of clusters. ").defaultValue("").required(true)
79 descriptor = training_data_path :: descriptor
80 descriptor = model_save_path :: descriptor
81 descriptor = maxIter :: descriptor
82 descriptor = minTol :: descriptor
83 descriptor = k :: descriptor
84 descriptor
85 }
86
87 override def getIcon(): Array[Byte] = {
88 ImageUtil.getImage("icon/ml_clustering/KmeansTraining.png")
89 }
90
91 override def getGroup(): List[String] = {
92 List(StopGroup.MLGroup.toString)
93 }
94
95}