2021-10-31

ALS算法 java spark
import org.apache.spark.SparkConf;import org.apache.spark.api.java.JavaPairRDD;import org.apache.spark.api.java.JavaRDD;import org.apache.spark.api.java.JavaSparkContext;import org.apache.spark.mllib.recommendation.ALS;import org.apache.spark.mllib.recommendation.MatrixFactorizationModel;import org.apache.spark.mllib.recommendation.Rating;import scala.Tuple2;public class myAls { public static void main(String[] args) { // TODO Auto-generated method stubSparkConf conf=new SparkConf().setAppName(“als”).setMaster(“local”);JavaSparkContext sc=new JavaSparkContext(conf);JavaRDD con=sc.textFile(“file:///home/gyq/下载/spark-2.3.2-bin-hadoop2.7/data/mllib/als/sample.data”);JavaRDD ratings=con.map(f->{ return new Rating( new Integer(f.split(":[0]), new Integer(f.split(":[1]), new Double(f.split(":[2]));});//数据转换为javardd三元组JavaRDD[] rr=ratings.randomSplit(new double[]{0.3,0.7});MatrixFactorizationModel model=ALS.train(rr[0].rdd(), 15, 10);//生成模型JavaRDD> up =rr[1].map(f->{//取0.7的数据要预测的用户和产品ID return new Tuple2<>(f.user(),f.product());});JavaPairRDD pupr= model.predict(up.rdd()).toJavaRDD().mapToPair(f->{ return new Tuple2(f.user()+","+f.product(),f.rating());//用0.7的数据去预测得到一个键值对});//pupr.foreach(f->System.out.println(f+“gg”));JavaPairRDD upr=rr[1].mapToPair(f->{//原来0.3的数据转化为键值对 return new Tuple2(f.user()+","+f.product(),f.rating());});//upr.foreach(f->System.out.println(f+“ss”));JavaPairRDD> mm=upr.join(pupr);//mm.foreach(f->System.out.println(f+“qq”));//model.save(sc.sc(),“file:///home/gyq/下载/spark-2.3.2-bin-hadoop2.7/data/mllib/als/myals”);JavaPairRDD> recom=mm.filter(f->{ double f2_2=f._2._2; double f2_1=f._2._1; int a=(int)f2_1; int b=(int)f2_2; if (a==b){ return true;} else return false;});double count=recom.count();double counts=mm.count();double accur=count/counts;System.out.println(“count=”+count);System.out.println(“counts=”+counts);System.out.println(“accur=”+accur);sc.stop();