双色球预测的一次尝试——前期准备与初步实施



萌发预测双色球的想法来源于很久以前，小时候电视上看双色球开奖结果的时候，总会发现当解说员说“下一个中奖号码是···”的时候，中奖号码就跳出来了，所以我认为是可以认为控制的，至于可不可以预测，那就不知道了，反正试试吧。

首先我得收集双色球的中奖号码，通过历史中奖号码来预测下一组号码。

表结构设计：

中奖号码表：

字段名	名称	类型	备注
lottery_date	日期	varchar	主键
phase	期号	varchar
red_1	1号球	int
red_2	2号球	int
red_3	3号球	int
red_4	4号球	int
red_5	5号球	int
red_6	6号球	int
blue	蓝色球	int
lucky_blue	幸运蓝球	int

预测表：

字段名	名称	类型	备注
id	ID	int	主键
phase	期号	varchar
red_1	1号球	int
red_2	2号球	int
red_3	3号球	int
red_4	4号球	int
red_5	5号球	int
red_6	6号球	int
blue	蓝色球	int
lucky_blue	幸运蓝球	int

误差表：

字段名	名称	类型	备注
id	ID	int	主键
phase	期号	varchar
red_1	1号球	int
red_2	2号球	int
red_3	3号球	int
red_4	4号球	int
red_5	5号球	int
red_6	6号球	int
blue	蓝色球	int
lucky_blue	幸运蓝球	int

收集数据：用Java编写爬虫收集双色球中奖数据，顺便说下爬虫的使用时我在超级课程表面试的时候学会使用的，现在可以用到这个上面来实在是太幸运了。

Java代码爬虫部分

/**
 * this class will grab the information of each lottery
 * run this method you will grab the latest information of the lottery
 * 
 *
 */

public class Spider {

	public static String getHtmlByUrl(String url) {
		String html = null;
		HttpClient httpClient = new DefaultHttpClient();// 创建httpClient对象
		HttpGet httpget = new HttpGet(url);// 以get方式请求该URL
		try {
			HttpResponse responce = httpClient.execute(httpget);// 得到responce对象
			int resStatu = responce.getStatusLine().getStatusCode();// 返回码
			if (resStatu == HttpStatus.SC_OK) {// 200正常 其他就不对
				// 获得相应实体
				HttpEntity entity = responce.getEntity();
				if (entity != null) {
					html = EntityUtils.toString(entity);// 获得html源代码
				}
			}
		} catch (Exception e) {
			System.out.println("访问【" + url + "】出现异常!");
			e.printStackTrace();
		} finally {
			httpClient.getConnectionManager().shutdown();
		}
		return html;
	}

	/**
	 * @param args
	 */
	public static void main(String[] args) {

		DataStore dataStore = new DataStore();
		Lottery lottery = new Lottery();

		String html = getHtmlByUrl("http://baidu.lecai.com/lottery/draw/list/ssq.php");
		if (html != null && !"".equals(html)) {
			Document doc = Jsoup.parse(html);
			Elements trs = doc.select("table").first().select("tbody").select("tr");
			Elements tds = trs.get(4).select("td");

			String phase = tds.get(0).text();
			lottery.setPhase(phase);
			
			String date = tds.get(1).text();
			date = date.substring(0, 10);
			lottery.setDate(date);
			
			Elements tds2 = tds.get(2).select("em");
			int[] a = new int[tds2.size()];
			for (int i = 0; i < tds2.size(); i++) {
				
				String text = tds2.get(i).text();
				a[i] = Integer.parseInt(text);
			}
			
			lottery.setRed_1(a[0]);
			lottery.setRed_2(a[1]);
			lottery.setRed_3(a[2]);
			lottery.setRed_4(a[3]);
			lottery.setRed_5(a[4]);
			lottery.setRed_6(a[5]);
			lottery.setBlue(a[6]);
			if(a.length == 8){
				lottery.setLucky_blue(a[a.length-1]);
			}
			dataStore.storeData(lottery);

		}
	}

}

数据存储部分，使用了Hibernate：

public void storeData(Lottery lottery){
		
		System.out.println(lottery.getDate());
		System.out.println(lottery.getPhase());
		System.out.println(lottery.getRed_1());
		System.out.println(lottery.getRed_2());
		System.out.println(lottery.getRed_3());
		System.out.println(lottery.getRed_4());
		System.out.println(lottery.getRed_5());
		System.out.println(lottery.getRed_6());
		System.out.println(lottery.getBlue());
		System.out.println(lottery.getLucky_blue());
		try{
			session=HibernateUtils.getSession();
			session.beginTransaction();
			session.save(lottery);
			
			session.getTransaction().commit();
			System.out.println("the date stored successfully");
		}catch(Exception e) {
			e.printStackTrace();
			session.getTransaction().rollback();
		}finally {
			HibernateUtils.closeSession(session);
		}
		
		
	}

缺点是如果网页结构改变的话，代码也需要重新修改。

目前我收集是从2014年10月19日到2015年4月7日的数据。

 

画出每个号码的走势图：使用Python。

import mysql.connector
from matplotlib.pyplot import plot
from matplotlib.pyplot import show

conn = mysql.connector.connect(user='root', password='root', database='lottery_forecast', use_unicode=True)
cursor = conn.cursor()

cursor.execute('select * from tb_lottery')
values = cursor.fetchall()

phase = []
red1 = []
red2 = []
red3 = []
red4 = []
red5 = []
red6 = []
blue = []

for i in values:
    #print i[2:9]
    phase.append(i[1])
    red1.append(i[2])
    red2.append(i[3])
    red3.append(i[4])
    red4.append(i[5])
    red5.append(i[6])
    red6.append(i[7])
    blue.append(i[8])
    
plot(red1,lw=1.0)
plot(red2,lw=1.0)
plot(red3,lw=1.0)
plot(red4,lw=1.0)
plot(red5,lw=1.0)
plot(red6,lw=1.0)
plot(blue,lw=2.0)
show()

结果如下图：

细线：深蓝色：1号球，绿色：2号球，红色：3号球，淡蓝色：4号球，紫色：5号球，黄色：6号球。

黑色粗线：蓝球

大致看上去还有点规律样的···

 

预测：Python代码：把每个球的历史中奖号码传入。

import numpy as np
import sys

def prediction( num):
    N = len(num)

    b = num[-N:]
    b = b[::-1]

    A = np.zeros((N, N), int)

    for i in range(N):
        A[i,]=num
        num1=[]
        num1.append(num[N-1])
        for j in range(N-1):
            num1.append(num[j])
        num=num1

    (x, residuals, rank, s) = np.linalg.lstsq(A,b)
    # print x, residuals, rank, s
    return np.dot(b,x)

2015年4月6日的计算结果是2015039期：1, 2, 12 , 22, 28, 29, 9，实际开奖结果：1, 13, 15, 26, 29, 30，12。中两个号码···

 

误差计算：Java代码

/**
 * when you did your prediction
 * and the real numbers come out
 * you can run this method to figure out the error
 * so you can make the next prediction according to this error
 * 
 *
 */

public class ErrorCalculate {
	
	
	private static Session session;
	Lottery lottery = new Lottery();
	Forecast forecast = new Forecast();
	ForecastError error = new ForecastError();
	
	public void errorCalculate(){
		
		//query the latest lottery information
		List lotteryList = null;
		
		try{
			session=HibernateUtils.getSession(); 
			session.beginTransaction();         
			Query query = session.createQuery("from Lottery l where 1=1 order by l.date desc");
			query.setFirstResult(0);
			query.setMaxResults(1);
			lotteryList = query.list();     
			session.getTransaction().commit(); 
		}catch(Exception e) {
			e.printStackTrace();
			session.getTransaction().rollback();
		}finally {
			HibernateUtils.closeSession(session);
		}
		
		for (int i = 0; i < lotteryList.size(); i++) {
			lottery = (Lottery) lotteryList.get(i);
			
			//show the lottery's phase to make sure that the two phases are the same
			System.out.println(lottery.getPhase());
		}
		
		
		//query the latest forecasted information
		List forecastList = null;
		try{
			session=HibernateUtils.getSession(); 
			session.beginTransaction();         
			Query query = session.createQuery("from Forecast f where 1=1 order by f.phase desc");
			query.setFirstResult(0);
			query.setMaxResults(1);
			forecastList = query.list();     
			session.getTransaction().commit(); 
		}catch(Exception e) {
			e.printStackTrace();
			session.getTransaction().rollback();
		}finally {
			HibernateUtils.closeSession(session);
		}
		
		for (int i = 0; i < forecastList.size(); i++) {
			forecast = (Forecast) forecastList.get(i);
			
			//show the forecast phase to make sure that the two phase are the same
			System.out.println(forecast.getPhase());
		}
		
		//The error calculate must be made while the latest lottery's phase is the same to the latest forecast's phase
		if(lottery.getPhase().equals(forecast.getPhase())){
			//calculate the error by the formula error = lottery - forecast
			error.setPhase(lottery.getPhase());  //use the lottery and the forecast phase as the phase of error
			error.setRed_1(lottery.getRed_1() - forecast.getRed_1());
			error.setRed_2(lottery.getRed_2() - forecast.getRed_2());
			error.setRed_3(lottery.getRed_3() - forecast.getRed_3());
			error.setRed_4(lottery.getRed_4() - forecast.getRed_4());
			error.setRed_5(lottery.getRed_5() - forecast.getRed_5());
			error.setRed_6(lottery.getRed_6() - forecast.getRed_6());
			error.setBlue(lottery.getBlue() - forecast.getBlue());
			//error.setLucky_blue(lottery.getLucky_blue() - forecast.getLucky_blue());
			
			
			System.out.println(error.getRed_1());
			System.out.println(error.getRed_2());
			System.out.println(error.getRed_3());
			System.out.println(error.getRed_4());
			System.out.println(error.getRed_5());
			System.out.println(error.getRed_6());
			System.out.println(error.getBlue());
			//System.out.println(error.getLucky_blue());
			
			//save the error into table error
			try{
				session=HibernateUtils.getSession();
				session.beginTransaction();
				session.save(error);
				session.getTransaction().commit();
				System.out.println("the error stored successfully");
			}catch(Exception e) {
				e.printStackTrace();
				session.getTransaction().rollback();
			}finally {
				HibernateUtils.closeSession(session);
			}
		}else{
			System.out.println("the phases of lottery and forecast are not same, please make the forecast or wait the next lottery comes out.");
		}
		
	}

2015年4月8日，再次运行Python预测算法，得到结果：1, 2, 13, 22, 28, 30, 9，怎么跟上次是差不多的啊，下次学着用其他方式试试。