xgboost算法预测彩票中奖号

仅供学习
效果图：

步骤：
1.先爬取排5历史的开奖记录，保存至CSV文件
2.将数据保存成dataset.pkl文件，方便日后追加排3的中奖号码
3.使用近128期的历史出奖记录建立序列模型，预测下期的出奖号
4.更新dataset.pkl文件，方便3步骤调用

1.爬虫：data.py

# -*- coding:utf-8 -*-

import time
import requests
from lxml import etree
import re
import os
import logging
import json
import csv
logging.basicConfig(format="%(asctime)s %(message)s",datefmt="%Y-%m-%d %I:%M:%S %p")   #格式化日志

class My_Data:
    
    def __init__(self,urls,data,header):
        self.urls = urls
        self.data = data
        self.header = header

    def it_url(self,count,n):    
        """
        定义一个生成器，用于生成每页url的data参数
        """
        
        for i in range(1,count+1,n):
            data["pageNo"] = i
            yield data   
        
    def page_count(self):
        """子url数"""
        try:
            counts = 192
            n = 1
            return counts,n
            
        except Exception as e:
            logging.WARNING("请求访问失败",e)
            return None
    
    def children_url(self):
        """获取html"""
        count,n = self.page_count()
        try:
            for dt in self.it_url(count,n):
                response = requests.get(self.urls,headers=self.header,params=dt)
                #print(response.url)    #用于校验url是否正确
                if response.status_code == requests.codes.ok:
                    html = response.text
                    logging.warning(f"第{k}请求成功,URL为{response.url}")
                    self.parsing(html)   #调用解析html方法  
        except:
            return None

    def parsing(self,html):
        """用于解析html"""
        datas = json.loads(html)
        num = len(datas["value"]["list"])
        filename = "pai5_date.csv"
        with open(filename,"a",newline="") as csvfile:
            writer = csv.writer(csvfile)
            for i in range(num):
                time = datas["value"]["list"][i]["lotteryDrawTime"]
                values = datas["value"]["list"][i]["lotteryDrawResult"]
                values = (time,values)
                writer.writerow(values)
                
if __name__ == "__main__":
    k = 1 
    header = {"user-agent":"Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko)\
    Chrome/84.0.4147.105 Safari/537.36"}
    data = {"pageNo":0}
    urls = "https://webapi.sporttery.cn/gateway/lottery/getHistoryPageListV1.qry?gameNo=350133&provinceId=0&pageSize=30&isVerify=1"  
    m = My_Data(urls,data,header)
    m.children_url()

2.预处理：pre.dataset.py

# -*- coding:utf-8 -*-

import numpy as np
import pandas as pd
from joblib import load,dump
from matplotlib import pyplot as plt
plt.rcParams['font.sans-serif'] = ['SimHei']  # 用来正常显示中文标签
plt.rcParams['axes.unicode_minus'] = False  # 用来正常显示负号

#第一次处理数据
#pf = pd.read_csv("pai5_date.csv",header=None)
##拆分每个数据
#data = []
#for i in range(len(pf[1])):
    #data.append(pf[1][i].split())
#pff = pd.DataFrame(data)
#dump(pff,r"model\dataset.pkl")

#导入原数据
pffs = load(r"model\dataset.pkl")

#获取最近一期时间
t = pffs["time"].tolist()
print(f"最近一期数据的时间为:{t[0]}")

#输入进几期排5数据，返回datafrom形式
n = int(input("请输入输入数据的组数:"))
data_list = []
time = []
for i in range(n):
    time = input("请输入开奖日期，如2021-01-23:")
    data_list.append(input("请输入开奖号码，用空格割开，如：8  4  0  6  7:").split())
#转换数据为array形式
data_arr = np.array(data_list).astype(float)
time_arr = np.array(time).reshape(-1,1)
#转换成dataframe形式
data_pd = pd.DataFrame(data_arr)
time_pd = pd.DataFrame(time_arr)
#合并数据
data_pd.insert(0,"time",time_pd)


#合并现在数据与原先数据
pff = pd.concat([data_pd,pffs])
dump(pff,r"model\dataset.pkl")

3.序列模型预测

# -*- coding:utf-8 -*-
from matplotlib import pyplot as plt
import pandas as pd 
import numpy as np 
import xgboost as xgb
from sklearn.model_selection import train_test_split
from sklearn.metrics import mean_squared_error
plt.rcParams['font.sans-serif'] = ['SimHei']  # 用来正常显示中文标签
plt.rcParams['axes.unicode_minus'] = False  # 用来正常显示负号
from joblib import dump,load

def dataset(a1):
    x,y = [],[]
    for i in range(n,len(a1)-1):
        x.append(a1[i-n:i])
        y.append(a1[i+1])
    x = np.array(x).astype(float)
    y = np.array(y).astype(float).T
    return x,y

def train(X_train, y_train):
    model = xgb.XGBRegressor(max_depth=20, learning_rate=0.1, n_estimators=50, silent=True, objective='reg:gamma')
    #训练
    model.fit(X_train, y_train)
    return model

if __name__ == "__main__":
    pff = load(r"model\dataset.pkl")
    #设置维度数
    n = 128     
    aa = pff[0].tolist() 
    bb = pff[1].tolist()
    cc = pff[2].tolist()
    aa = aa[::-1]
    bb = bb[::-1]
    cc = cc[::-1]    
    x1s,y1s = dataset(aa)
    x2s,y2s = dataset(bb)
    x3s,y3s = dataset(cc)
    #划分数据
    X1_train,X1_test,y1_train,y1_test = train_test_split(x1s[:len(x1s)-1],y1s[:len(x1s)-1],test_size=0.2,random_state=1234)
    X2_train,X2_test,y2_train,y2_test = train_test_split(x2s[:len(x1s)-1],y2s[:len(x1s)-1],test_size=0.2,random_state=1234)
    X3_train,X3_test,y3_train,y3_test = train_test_split(x3s[:len(x1s)-1],y3s[:len(x1s)-1],test_size=0.2,random_state=1234)
        
    #训练
    model1 = train(X1_train,y1_train)
    model2 = train(X2_train,y2_train)
    model3 = train(X3_train,y3_train)
    #预测
    x1 = aa[len(aa)-n-1:len(aa)-1]
    x1 = np.array(x1).astype(float)
    x1 = np.expand_dims(x1, 0)
    
    x2 = bb[len(bb)-n-1:len(bb)-1]
    x2 = np.array(x2).astype(float)
    x2 = np.expand_dims(x2, 0)  #增加一个维度
    
    x3 = cc[len(cc)-n-1:len(cc)-1]
    x3 = np.array(x3).astype(float)
    x3 = np.expand_dims(x3, 0)


    yy_predict1 = model1.predict(x1)
    yy_predict2 = model2.predict(x2)
    yy_predict3 = model3.predict(x3)
    print(f"{np.around(yy_predict1[0],4)} -- {np.around(yy_predict2[0],4)} -- {np.around(yy_predict3[0],4)}")

    #画图
    totel = len(y1s)
    xx1 = x1s[-1:totel-100:-1]
    yy1 = y1s[-1:totel-100:-1]    
    yy_pred1 = model1.predict(xx1)
    yy1 = yy1[-1::-1]  #方便画图
    yy_pred1 = yy_pred1[-1::-1]
    print(f"预测值:{yy_pred1[-1]},真实值:{yy1[-1]}")
    
    xx2 = x2s[-1:totel-100:-1]
    yy2 = y2s[-1:totel-100:-1]    
    yy_pred2 = model2.predict(xx2)
    yy2 = yy2[-1::-1]  #方便画图
    yy_pred2 = yy_pred2[-1::-1]
    print(f"预测值:{yy_pred2[-1]},真实值:{yy2[-1]}")
    
    xx3 = x3s[-1:totel-100:-1]
    yy3 = y3s[-1:totel-100:-1]    
    yy_pred3 = model3.predict(xx3)
    yy3 = yy3[-1::-1]  #方便画图
    yy_pred3 = yy_pred3[-1::-1]
    print(f"预测值:{yy_pred3[-1]},真实值:{yy3[-1]}")
    #准确率
    acc = (yy1 ==np.around(yy_pred1)).sum()/len(yy1)
    
    plt.figure(figsize=(16,12))
    plt.subplot(311)
    plt.plot(list(range(len(yy1))),yy1,c="r",label="真实值")
    plt.plot(list(range(len(yy1))),yy_pred1,c="b",label="预测值")
    plt.legend(loc='upper left')
    plt.text(40, 8, f"1号位置：上期预测值:{yy_pred1[-1]},真实值:{yy1[-1]}")
    plt.text(87, yy_predict1[0],f"预测值：{round(yy_predict1[0],4)}")
    
    plt.subplot(312)
    plt.plot(list(range(len(yy1))),yy2,c="r",label="真实值")
    plt.plot(list(range(len(yy1))),yy_pred2,c="b",label="预测值")
    plt.legend(loc='upper left')
    plt.text(40, 8, f"2号位置：上期预测值:{yy_pred2[-1]},真实值:{yy2[-1]}")
    plt.text(87, yy_predict2[0]+1,f"预测值：{round(yy_predict2[0],4)}")
    
    plt.subplot(313)
    plt.plot(list(range(len(yy1))),yy3,c="r",label="真实值")
    plt.plot(list(range(len(yy1))),yy_pred3,c="b",label="预测值")
    plt.legend(loc='upper left')
    plt.text(40, 8, f"3号位置：上期预测值:{yy_pred3[-1]},真实值:{yy3[-1]}")
    plt.text(87, yy_predict3[0],f"预测值：{round(yy_predict3[0],4)}")
    plt.savefig("排3预测图.png")
    #plt.show()

注：纸上得来终觉浅，绝知此事要躬行。