Mahout使用(一)

1.HelloMahout.java
2.DistanceTest.java
3.MahoutDemo.java

1.HelloMahout.java

package cn.crxy.mahout;

import java.io.File;
import java.util.List;

import org.apache.log4j.Logger;
import org.apache.mahout.cf.taste.impl.model.file.FileDataModel;
import org.apache.mahout.cf.taste.impl.neighborhood.NearestNUserNeighborhood;
import org.apache.mahout.cf.taste.impl.recommender.GenericUserBasedRecommender;
import org.apache.mahout.cf.taste.impl.similarity.PearsonCorrelationSimilarity;
import org.apache.mahout.cf.taste.model.DataModel;
import org.apache.mahout.cf.taste.neighborhood.UserNeighborhood;
import org.apache.mahout.cf.taste.recommender.RecommendedItem;
import org.apache.mahout.cf.taste.recommender.Recommender;
import org.apache.mahout.cf.taste.similarity.UserSimilarity;

public class HelloMahout {

    public static void main(String[] args) {
        
        Logger logger=Logger.getLogger(HelloMahout.class);
        try {
            //读取用户评分数据    封装成一个model
            DataModel model = new FileDataModel(new File("F:\360Downloads\超人学院\第14期视频\2016-09-12【mahout】\样本数据\info.csv"));
            // 根据相似度找出对应的好朋友的标准     物以类聚,人以群分
            UserSimilarity userSimilarity = new PearsonCorrelationSimilarity(model);
            // 邻域 选择两个好朋友帮我推荐
            UserNeighborhood userNeighborhood = new NearestNUserNeighborhood(2,userSimilarity, model);
            // 构建推荐引擎
            Recommender recommender = new GenericUserBasedRecommender(model,userNeighborhood, userSimilarity);
            // 进行推荐
            List<RecommendedItem> recommend = recommender.recommend(1, 5);
            for (RecommendedItem item : recommend) {
                logger.info(item);
            }
        } catch (Exception e) {
            logger.error(e.getMessage());
        }
    }
}

2.DistanceTest.java

package cn.crxy.mahout;

import org.junit.Before;
import org.junit.Test;

public class DistanceTest {
    
    //    水果维度依次为：苹果、梨、桃子、栗子、香蕉、橘子
    //    小明：5,4,2,1,5,5
    //    小丽：5,3,1,2,1,1
    //    小王：5,3,4,1,4,3
    private int[] a;
    private int[] b;
    private int[] c;
    
    @Before
    public void initData(){
        a=new int[]{5,4,2,1,5,5};
        b=new int[]{5,3,1,2,1,1};
        c=new int[]{5,3,4,1,4,3};
    }
    
    @Test
    public void Distance(){
//        a-b:5.916079783099616
//        a-c:3.1622776601683795
//        c-b:4.795831523312719

        System.out.println(String.format("a-b:%s", 1.0/(1.0+Man(a, b))));
        System.out.println(String.format("a-c:%s", 1.0/(1.0+Man(a, c))));
        System.out.println(String.format("c-b:%s", 1.0/(1.0+Man(c, b))));
//        a-b:0.08333333333333333
//        a-c:0.14285714285714285
//        c-b:0.1
        
    }
    //欧式距离
    private double ErluD(int[] a_array,int[] b_array){
        double result=0;
        for (int i = 0; i < a_array.length; i++) {
            result+=Math.pow(a_array[i]-b_array[i],2);
        }
        return Math.sqrt(result);
    }
    //曼哈顿距离
    private double Man(int[] a_array,int[] b_array){
        double result=0;
        for (int i = 0; i < a_array.length; i++) {
            result+=Math.abs(a_array[i]-b_array[i]);
        }
        return result;
    }
    //min式距离
    private double Min(int[] a_array,int[] b_array,int p){
        double result=0;
        for (int i = 0; i < a_array.length; i++) {
            result+=Math.pow(Math.abs(a_array[i]-b_array[i]),p);
        }
        return Math.pow(result,1.0/p);
    }

}

3.MahoutDemo.java

package cn.crxy.mahout;

import java.io.File;
import java.util.List;

import org.apache.mahout.cf.taste.impl.common.FastByIDMap;
import org.apache.mahout.cf.taste.impl.common.FastIDSet;
import org.apache.mahout.cf.taste.impl.model.GenericPreference;
import org.apache.mahout.cf.taste.impl.model.GenericUserPreferenceArray;
import org.apache.mahout.cf.taste.impl.model.file.FileDataModel;
import org.apache.mahout.cf.taste.impl.neighborhood.NearestNUserNeighborhood;
import org.apache.mahout.cf.taste.impl.neighborhood.ThresholdUserNeighborhood;
import org.apache.mahout.cf.taste.impl.recommender.GenericBooleanPrefItemBasedRecommender;
import org.apache.mahout.cf.taste.impl.recommender.GenericItemBasedRecommender;
import org.apache.mahout.cf.taste.impl.recommender.GenericUserBasedRecommender;
import org.apache.mahout.cf.taste.impl.similarity.CachingItemSimilarity;
import org.apache.mahout.cf.taste.impl.similarity.CachingUserSimilarity;
import org.apache.mahout.cf.taste.impl.similarity.PearsonCorrelationSimilarity;
import org.apache.mahout.cf.taste.impl.similarity.TanimotoCoefficientSimilarity;
import org.apache.mahout.cf.taste.model.DataModel;
import org.apache.mahout.cf.taste.model.PreferenceArray;
import org.apache.mahout.cf.taste.neighborhood.UserNeighborhood;
import org.apache.mahout.cf.taste.recommender.RecommendedItem;
import org.apache.mahout.cf.taste.recommender.Recommender;
import org.apache.mahout.cf.taste.similarity.ItemSimilarity;
import org.apache.mahout.cf.taste.similarity.UserSimilarity;
import org.junit.Before;
import org.junit.Ignore;
import org.junit.Test;

public class MahoutDemo {

    //组装datamodel
    
    // userid itemid score
    // 101 102 103 104
    // 1(5,4,2,)
    // 2(,2,4,1)
    // 3(4,3,1,)
    DataModel dataModel;
    
    @Before
    public void initData() throws Exception{
        //每一个用户的喜好列表 key：用户id  value：该用户的偏好列表
        FastByIDMap<PreferenceArray> data=new FastByIDMap<PreferenceArray>();
        //组装第一个用户 偏好列表
        PreferenceArray array1=new GenericUserPreferenceArray(3);
        //PreferenceArray index 指：偏好列表的index 序号。
        array1.setUserID(0, 1);
        array1.setItemID(0, 101);
        array1.setValue(0, 5);
        
        array1.setUserID(1, 1);
        array1.setItemID(1, 102);
        array1.setValue(1, 4);
        
        array1.setUserID(2, 1);
        array1.setItemID(2, 103);
        array1.setValue(2, 2);
        
        data.put(1, array1);
        
        //组装第二个喜好
        PreferenceArray array2=new GenericUserPreferenceArray(3);
        //2(,2,4,1)
        array2.set(0, new GenericPreference(2,102,2));
        array2.set(1, new GenericPreference(2,103,4));
        array2.set(2, new GenericPreference(2,104,1));
        data.put(2, array2);
        //组装第三个喜好
        PreferenceArray array3=new GenericUserPreferenceArray(3);
        //3(4,3,1,)
        array3.set(0, new GenericPreference(3,101,4));
        array3.set(1, new GenericPreference(3,102,3));
        array3.set(2, new GenericPreference(3,103,1));
        data.put(3, array3);
        
        //dataModel=new GenericDataModel(data);
//        dataModel=new GenericBooleanPrefDataModel(userData);
//        System.out.println(dataModel.getPreferenceValue(1, 102));//获得1用户对102的评分
//        System.out.println(dataModel.getItemIDsFromUser(1));
//        System.out.println(dataModel.getUserIDs());
        
        
        //1  101 102 103
        //2 102 103 
        // key为userid value:物品的集合 set
        FastByIDMap<FastIDSet> userData=new FastByIDMap<FastIDSet>();
        
        FastIDSet userSet1=new FastIDSet(3);
        userSet1.add(101);
        userSet1.add(102);
        userSet1.add(103);
        userData.put(1,userSet1);
        
        FastIDSet userSet2=new FastIDSet(2);
        userSet2.add(102);
        userSet2.add(103);
        userData.put(2,userSet2);
        
        
        //无偏好的构建
//        dataModel=new GenericBooleanPrefDataModel(userData);
        
        
        //读取文件 有偏好的
        dataModel=new FileDataModel(new File("F:\360Downloads\超人学院\第14期视频\2016-09-12【mahout】\样本数据\info.csv"));
        //读取文件 无偏好的 无偏好的数据只有用户和其关联的商品 没有对应商品的评分
//        dataModel=new FileDataModel(new File("F:\360Downloads\超人学院\第14期视频\2016-09-12【mahout】\样本数据\ubool.data"));
        
        
//        对于无偏好数据：getvalue：如果存在记录则是1.0；否则为null。  
//        System.out.println(dataModel.getPreferenceValue(1, 103));
//        System.out.println(dataModel.getItemIDsFromUser(1));
//        System.out.println(dataModel.getUserIDs());
        
    }
    @Ignore
    public void testUserSimi() throws Exception{
        
        //利用model和相似度函数 计算用户相似度
//        UserSimilarity userSimilarity=new TanimotoCoefficientSimilarity(dataModel);
        UserSimilarity userSimilarity=new PearsonCorrelationSimilarity(dataModel);
        userSimilarity=new CachingUserSimilarity(userSimilarity, dataModel);
        //查询用户之间的相似度  0.9999999999999998    0.944911182523068
        //如果使用CachingUserSimilarity userSimilarity(1,5) 第二次不会再次计算了
        System.out.println(userSimilarity.userSimilarity(1, 5));
        System.out.println(userSimilarity.userSimilarity(1, 5));
    }
    @Ignore
    public void testItemSimi() throws Exception{
        
        //利用model和相似度函数 计算物品相似度
        ItemSimilarity itemSimilarity=new PearsonCorrelationSimilarity(dataModel);
        itemSimilarity =new CachingItemSimilarity(itemSimilarity,dataModel);
        //查询物品之间的相似度 0.9449111825230729
        System.out.println(itemSimilarity.itemSimilarity(101, 102));
    }
    @Test
    public void testuserNeighborhood() throws Exception{
        //相似度   有相似度才能算邻居是谁
        UserSimilarity userSimilarity=new PearsonCorrelationSimilarity(dataModel);
        //1.固定数目的邻居  如果取邻居 只取前三个 
        UserNeighborhood userNeighborhood=new NearestNUserNeighborhood(3,userSimilarity,dataModel);
        long[] userNeighborhoods = userNeighborhood.getUserNeighborhood(1);//为1用户取得用户
        for (long l : userNeighborhoods) {
            System.out.println(l+"NearestNUserNeighborhoodsimi---"+userSimilarity.userSimilarity(1, l));
        }
//        4NearestNUserNeighborhoodsimi---0.9999999999999998
//        5NearestNUserNeighborhoodsimi---0.944911182523068
//        2NearestNUserNeighborhoodsimi---   -0.7642652566278799这个是负0.7

    
        //2.固定阈值的邻居  只要0.8以上的
        userNeighborhood=new ThresholdUserNeighborhood(0.7,userSimilarity,dataModel);
        long[] userNeighborhoodsnew = userNeighborhood.getUserNeighborhood(1);
        System.out.println(userSimilarity.userSimilarity(1, 2)); //查看1和2的相似度
        for (long l : userNeighborhoodsnew) {
            System.out.println(l+"ThresholdUserNeighborhoodsimi---"+userSimilarity.userSimilarity(1, l));
        }
        
    }
    @Test
    public void testItemCmd() throws Exception{
        //1.基于物品的有偏好的推荐   基于物品的不需要邻居
//        ItemSimilarity itemSimilarity=new PearsonCorrelationSimilarity(dataModel);
//        Recommender recommender=new GenericItemBasedRecommender(dataModel,itemSimilarity);
        
        //2.基于物品的无偏好推荐
        ItemSimilarity itemSimilarity=new TanimotoCoefficientSimilarity(dataModel);
        Recommender recommender=new GenericBooleanPrefItemBasedRecommender(dataModel,itemSimilarity);
        
        
        List<RecommendedItem> recommend = recommender.recommend(1, 3);//给用户1推荐3个.
        for (RecommendedItem recommendedItem : recommend) {
            System.out.println(recommendedItem);
            //1.基于物品的有偏好的推荐RecommendedItem[item:104, value:5.0]其他的推荐不出来了....所以只推荐出了1个
            
            //2.基于物品的无偏好的推荐
            //RecommendedItem[item:104, value:1.8]
            //RecommendedItem[item:106, value:1.15]
            //RecommendedItem[item:105, value:0.85]
        }
    }
    @Test
    public void testUserCmd() throws Exception{
        //1.基于用户的有偏好的推荐
        //UserSimilarity userSimilarity=new PearsonCorrelationSimilarity(dataModel);
        //2.基于用户的无偏好的推荐
        UserSimilarity userSimilarity=new TanimotoCoefficientSimilarity(dataModel);
        
        UserNeighborhood userNeighborhood=new NearestNUserNeighborhood(3,userSimilarity,dataModel);//Top 3
        //构建推荐对象
        Recommender recommender=new GenericUserBasedRecommender(dataModel,userNeighborhood,userSimilarity);
        List<RecommendedItem> recommend = recommender.recommend(1, 3);
        for (RecommendedItem recommendedItem : recommend) {
            System.out.println(recommendedItem);
            //1.基于用户的有偏好推荐
            //RecommendedItem[item:104, value:5.0]
            //RecommendedItem[item:106, value:4.0]
            //2.基于用户的无偏好推荐
            //RecommendedItem[item:106, value:4.0]
            //RecommendedItem[item:104, value:3.2121212]

        }
    }
    
    
}