import java.sql.*;
import java.util.*;

/**
 * @author tianshl
 * @version 2018/1/13 上午11:13
 */
public class Kmeans {
  // 源數(shù)據(jù)
  private List<Integer> origins = new ArrayList<>();

  // 分組數(shù)據(jù)
  private Map<Double, List<Integer>> grouped;

  // 初始質(zhì)心列表
  private List<Double> cores;

  // 數(shù)據(jù)源
  private String tableName;
  private String colName;

  /**
   * 構(gòu)造方法
   *
   * @param tableName 源數(shù)據(jù)表名稱
   * @param colName  源數(shù)據(jù)列名稱
   * @param cores   質(zhì)心列表
   */
  private Kmeans(String tableName, String colName,List<Double> cores){
    this.cores = cores;
    this.tableName = tableName;
    this.colName = colName;
  }

  /**
   * 重新計(jì)算質(zhì)心
   *
   * @return 新的質(zhì)心列表
   */
  private List<Double> newCores(){
    List<Double> newCores = new ArrayList<>();

    for(List<Integer> v: grouped.values()){
      newCores.add(v.stream().reduce(0, (sum, num) -> sum + num) / (v.size() + 0.0));
    }

    Collections.sort(newCores);
    return newCores;
  }

  /**
   * 判斷是否結(jié)束
   *
   * @return bool
   */
  private Boolean isOver(){
    List<Double> _cores = newCores();
    for(int i=0, len=cores.size(); i<len; i++){
      if(!cores.get(i).toString().equals(_cores.get(i).toString())){
        // 使用新質(zhì)心
        cores = _cores;
        return false;
      }
    }
    return true;
  }

  /**
   * 數(shù)據(jù)分組
   */
  private void setGrouped(){
    grouped = new HashMap<>();

    Double core;
    for (Integer origin: origins) {
      core = getCore(origin);

      if (!grouped.containsKey(core)) {
        grouped.put(core, new ArrayList<>());
      }

      grouped.get(core).add(origin);
    }
  }

  /**
   * 選擇質(zhì)心
   *
   * @param num  要分組的數(shù)據(jù)
   * @return   質(zhì)心
   */
  private Double getCore(Integer num){

    // 差 列表
    List<Double> diffs = new ArrayList<>();

    // 計(jì)算差
    for(Double core: cores){
      diffs.add(Math.abs(num - core));
    }

    // 最小差 -> 索引 -> 對(duì)應(yīng)的質(zhì)心
    return cores.get(diffs.indexOf(Collections.min(diffs)));
  }

  /**
   * 建立數(shù)據(jù)庫(kù)連接
   * @return connection
   */
  private Connection getConn(){
    try {
      // URL指向要訪問(wèn)的數(shù)據(jù)庫(kù)名mydata
      String url = "jdbc:mysql://localhost:3306/data_analysis_dev";
      // MySQL配置時(shí)的用戶名
      String user = "root";
      // MySQL配置時(shí)的密碼
      String password = "root";

      // 加載驅(qū)動(dòng)
      Class.forName("com.mysql.jdbc.Driver");

      //聲明Connection對(duì)象
      Connection conn = DriverManager.getConnection(url, user, password);

      if(conn.isClosed()){
        System.out.println("連接數(shù)據(jù)庫(kù)失敗!");
        return null;
      }
      System.out.println("連接數(shù)據(jù)庫(kù)成功!");

      return conn;

    } catch (Exception e) {
      System.out.println("連接數(shù)據(jù)庫(kù)失??！");
      e.printStackTrace();
    }

    return null;
  }

  /**
   * 關(guān)閉數(shù)據(jù)庫(kù)連接
   *
   * @param conn 連接
   */
  private void close(Connection conn){
    try {
      if(conn != null && !conn.isClosed()) conn.close();
    } catch (Exception e){
      e.printStackTrace();
    }
  }

  /**
   * 獲取源數(shù)據(jù)
   */
  private void getOrigins(){

    Connection conn = null;
    try {
      conn = getConn();
      if(conn == null) return;

      Statement statement = conn.createStatement();

      ResultSet rs = statement.executeQuery(String.format("select %s from %s", colName, tableName));

      while(rs.next()){
        origins.add(rs.getInt(1));
      }
      conn.close();
    } catch (Exception e){
      e.printStackTrace();
    } finally {
     close(conn);
    }
  }

  /**
   * 向新表中寫數(shù)據(jù)
   */
  private void write(){

    Connection conn = null;
    try {
      conn = getConn();
      if(conn == null) return;
      
      // 創(chuàng)建表
      Statement statement = conn.createStatement();

      // 刪除舊數(shù)據(jù)表
      statement.execute("DROP TABLE IF EXISTS k_means; ");
      // 創(chuàng)建新表
      statement.execute("CREATE TABLE IF NOT EXISTS k_means(`core` DECIMAL(11, 7), `col` INTEGER(11));");

      // 禁止自動(dòng)提交
      conn.setAutoCommit(false);

      PreparedStatement ps = conn.prepareStatement("INSERT INTO k_means VALUES (?, ?)");

      for(Map.Entry<Double, List<Integer>> entry: grouped.entrySet()){
        Double core = entry.getKey();
        for(Integer value: entry.getValue()){
          ps.setDouble(1, core);
          ps.setInt(2, value);
          ps.addBatch();
        }
      }

      // 批量執(zhí)行
      ps.executeBatch();

      // 提交事務(wù)
      conn.commit();

      // 關(guān)閉連接
      conn.close();
    } catch (Exception e){
      e.printStackTrace();
    } finally {
      close(conn);
    }
  }

  /**
   * 處理數(shù)據(jù)
   */
  private void run(){
    System.out.println("獲取源數(shù)據(jù)");
    // 獲取源數(shù)據(jù)
    getOrigins();

    // 停止分組
    Boolean isOver = false;

    System.out.println("數(shù)據(jù)分組處理");
    while(!isOver) {
      // 數(shù)據(jù)分組
      setGrouped();
      // 判斷是否停止分組
      isOver = isOver();
    }

    System.out.println("將處理好的數(shù)據(jù)寫入數(shù)據(jù)庫(kù)");
    // 將分組數(shù)據(jù)寫入新表
    write();

    System.out.println("寫數(shù)據(jù)完畢");
  }

  public static void main(String[] args){
    List<Double> cores = new ArrayList<>();
    cores.add(260.0);
    cores.add(600.0);
    // 表名, 列名, 質(zhì)心列表
    new Kmeans("attributes", "attr_length", cores).run();
  }
}

Kmeans.java

javac Kmeans.java 

# 指定依賴庫(kù)
java -Djava.ext.dirs=./lib Kmeans