python进行大乐透和双色球选号（LSTM预测和随机选号）

文章仅供参考学习

1.LSTM预测

首先去爬取数据

这个是爬取大乐透的，从07年爬到最新一期

import requests
from bs4 import BeautifulSoup
import csv

# 目标URL
url = 'http://datachart.500.com/dlt/history/newinc/history.php?start=07001'

# 发送HTTP请求
response = requests.get(url)
response.encoding = 'utf-8'  # 确保编码正确

# 解析HTML内容
soup = BeautifulSoup(response.text, 'html.parser')

# 定位包含开奖数据的表格体
tbody = soup.find('tbody', id="tdata")

# 存储开奖数据的列表
lottery_data = []

# 遍历每一行数据
for tr in tbody.find_all('tr'):
    tds = tr.find_all('td')
    if tds:
        # 提取数据并添加到列表
        lottery_data.append([td.text for td in tds])

# 写入CSV文件
with open('dlt_lottery_data.csv', 'w', newline='', encoding='utf-8') as csvfile:
    writer = csv.writer(csvfile)
    # 写入标题行
    # writer.writerow(['期号', '号码1', '号码2', '号码3', '号码4', '号码5', '号码6', '号码7'])
    # 写入数据行
    writer.writerows(lottery_data)

print('数据抓取完成，并保存到dlt_lottery_data.csv文件中。')

下面是爬取双色球的

import requests
from bs4 import BeautifulSoup
import csv

# 目标URL
url = f'http://datachart.500.com/ssq/history/newinc/history.php?start=07001'

# 发送HTTP请求
response = requests.get(url)
response.encoding = 'utf-8'  # 确保编码正确

# 解析HTML内容
soup = BeautifulSoup(response.text, 'html.parser')

# 定位包含开奖数据的表格体
tbody = soup.find('tbody', id="tdata")

# 存储开奖数据的列表
lottery_data = []

# 遍历每一行数据
for tr in tbody.find_all('tr'):
    tds = tr.find_all('td')
    if tds:
        # 提取数据并添加到列表
        lottery_data.append([td.text for td in tds])

# 写入CSV文件
with open('ssq_lottery_data.csv', 'w', newline='', encoding='utf-8') as csvfile:
    writer = csv.writer(csvfile)
    # 写入标题行
    # writer.writerow(['期号', '号码1', '号码2', '号码3', '号码4', '号码5', '号码6', '号码7'])
    # 写入数据行
    writer.writerows(lottery_data)

print('数据抓取完成，并保存到ssq_lottery_data.csv文件中。')

对爬取的数据进行处理

大乐透是5+2，双色球是6+1，两个不同，注意区分。

大乐透的

import csv

import pandas as pd
def get_data(path):
    r_data = []
    b_data = []
    with open(path, 'r') as file:
        reader = csv.reader(file)
        for row in reader:
            r_data.append(list(map(lambda x: int(x), row[1:7])))
            b_data.append(list(map(lambda x: int(x), row[7:8])))
    r_data.reverse()
    b_data.reverse()
    return r_data, b_data

def process_data():
    p = r"./ssq_lottery_data.csv"
    r_data, b_data = get_data(p)
    # print(b_data)
    return r_data, b_data

if __name__ == '__main__':

    process_data()

下面是双色球的

import csv

import pandas as pd
def get_data(path):
    r_data = []
    b_data = []
    with open(path, 'r') as file:
        reader = csv.reader(file)
        for row in reader:
            r_data.append(list(map(lambda x: int(x), row[1:7])))
            b_data.append(list(map(lambda x: int(x), row[7:8])))
    r_data.reverse()
    b_data.reverse()
    return r_data, b_data

def process_data():
    p = r"./ssq_lottery_data.csv"
    r_data, b_data = get_data(p)
    # print(b_data)
    return r_data, b_data

if __name__ == '__main__':

    process_data()

下面开始定义模型

# 定义 LSTM 模型
class LSTMModel(nn.Module):
    def __init__(self, input_size, hidden_size, output_size, num_layers=1):
        super(LSTMModel, self).__init__()
        self.hidden_size = hidden_size
        self.num_layers = num_layers
        self.lstm = nn.LSTM(input_size, hidden_size, num_layers, batch_first=True)
        self.fc = nn.Linear(hidden_size, output_size)

    def forward(self, x):
        h0 = torch.zeros(self.num_layers, x.size(0), self.hidden_size).to(x.device)
        c0 = torch.zeros(self.num_layers, x.size(0), self.hidden_size).to(x.device)

        out, _ = self.lstm(x, (h0, c0))
        out = self.fc(out[:, -1, :])
        return out

对训练之前的数据进行标准化处理和转换成为tensor格式

def trans_process_data(seq_length):
    r_data, b_data = process_data()
    # print(r_data)

    # print(r_data)
    r_data = np.array(r_data)
    b_data = np.array(b_data)
    # 转换为 PyTorch 张量
    r_data = torch.tensor(r_data, dtype=torch.float32)
    # 转换为 PyTorch 张量
    b_data = torch.tensor(b_data, dtype=torch.float32)

    # 标准化
    r_mean = r_data.mean(dim=0)
    r_std = r_data.std(dim=0)
    r_data = (r_data - r_mean) / r_std

    # 标准化
    b_mean = b_data.mean(dim=0)
    b_std = b_data.std(dim=0)
    b_data = (b_data - b_mean) / b_std

    r_train = []
    r_target = []
    b_train = []
    b_target = []

    for i in range(len(r_data) - seq_length):
        r_train.append(r_data[i:i + seq_length])
        r_target.append(r_data[i + seq_length])
    r_train = torch.stack(r_train)
    r_target = torch.stack(r_target)

    for i in range(len(b_data) - seq_length):
        b_train.append(b_data[i:i + seq_length])
        b_target.append(b_data[i + seq_length])
    b_train = torch.stack(b_train)
    b_target = torch.stack(b_target)
    # print(r_train)
    return r_data, b_data, r_train, r_target, b_train, b_target, r_mean, r_std, b_mean, b_std

训练函数

def start_train(input_size, hidden_size, output_size, num_layers, train_data, target_data, num_epochs=100):
    model = LSTMModel(input_size, hidden_size, output_size, num_layers)
    criterion = nn.MSELoss()
    optimizer = optim.Adam(model.parameters(), lr=0.05)
    # 训练模型
    for epoch in range(num_epochs):
        model.train()
        optimizer.zero_grad()
        # 前向传播
        outputs = model(train_data)
        loss = criterion(outputs, target_data)
        # 反向传播和优化
        loss.backward()
        optimizer.step()
        if (epoch + 1) % 10 == 0:
            print(f'Epoch [{epoch + 1}/{num_epochs}], Loss: {loss.item():.4f}')
        if epoch == int(num_epochs/2):
            optimizer = optim.Adam(model.parameters(), lr=0.01)
    return model

预测函数

def start_predicted(model, predicted_data):
    model.eval()
    with torch.no_grad():
        test_input = predicted_data.unsqueeze(0)  # 使用最后seq_length个时间步作为输入
        predicted = model(test_input)
        # print("Predicted:", predicted)
    return predicted

红球和篮球分开训练预测，开始两个训练预测

def start_all_train(hidden_size, num_layers, num_epochs, seq_length):
    r_data, b_data, r_train, r_target, b_train, b_target, r_mean, r_std, b_mean, b_std = trans_process_data(seq_length)
    # print(r_mean, r_std)
    r_size = 5
    r_model = start_train(r_size, hidden_size, r_size, num_layers, r_train, r_target, num_epochs)
    predicted_data = r_data[-seq_length:]
    r_predicted = start_predicted(r_model, predicted_data)
    print("--------------------------bbbbb-------------------------------------------")
    b_size = 2
    b_model = start_train(b_size, hidden_size, b_size, num_layers, b_train, b_target, num_epochs)
    predicted_data = b_data[-seq_length:]
    b_predicted = start_predicted(b_model, predicted_data)

    print(r_predicted)
    print(b_predicted)

    r_predicted = r_predicted * r_std + r_mean
    b_predicted = b_predicted * b_std + b_mean

    print(r_predicted)
    print(b_predicted)

    return r_predicted, b_predicted

完整代码

import os
import sys

BASE_DIR = os.path.dirname(os.path.abspath(__file__))
sys.path.append(BASE_DIR)

from data_process import process_data
import torch
import torch.nn as nn
import torch.optim as optim
import numpy as np

# 定义 LSTM 模型
class LSTMModel(nn.Module):
    def __init__(self, input_size, hidden_size, output_size, num_layers=1):
        super(LSTMModel, self).__init__()
        self.hidden_size = hidden_size
        self.num_layers = num_layers
        self.lstm = nn.LSTM(input_size, hidden_size, num_layers, batch_first=True)
        self.fc = nn.Linear(hidden_size, output_size)

    def forward(self, x):
        h0 = torch.zeros(self.num_layers, x.size(0), self.hidden_size).to(x.device)
        c0 = torch.zeros(self.num_layers, x.size(0), self.hidden_size).to(x.device)

        out, _ = self.lstm(x, (h0, c0))
        out = self.fc(out[:, -1, :])
        return out

def trans_process_data(seq_length):
    r_data, b_data = process_data()
    # print(r_data)

    # print(r_data)
    r_data = np.array(r_data)
    b_data = np.array(b_data)
    # 转换为 PyTorch 张量
    r_data = torch.tensor(r_data, dtype=torch.float32)
    # 转换为 PyTorch 张量
    b_data = torch.tensor(b_data, dtype=torch.float32)

    # 标准化
    r_mean = r_data.mean(dim=0)
    r_std = r_data.std(dim=0)
    r_data = (r_data - r_mean) / r_std

    # 标准化
    b_mean = b_data.mean(dim=0)
    b_std = b_data.std(dim=0)
    b_data = (b_data - b_mean) / b_std

    r_train = []
    r_target = []
    b_train = []
    b_target = []

    for i in range(len(r_data) - seq_length):
        r_train.append(r_data[i:i + seq_length])
        r_target.append(r_data[i + seq_length])
    r_train = torch.stack(r_train)
    r_target = torch.stack(r_target)

    for i in range(len(b_data) - seq_length):
        b_train.append(b_data[i:i + seq_length])
        b_target.append(b_data[i + seq_length])
    b_train = torch.stack(b_train)
    b_target = torch.stack(b_target)
    # print(r_train)
    return r_data, b_data, r_train, r_target, b_train, b_target, r_mean, r_std, b_mean, b_std

def start_train(input_size, hidden_size, output_size, num_layers, train_data, target_data, num_epochs=100):
    model = LSTMModel(input_size, hidden_size, output_size, num_layers)
    criterion = nn.MSELoss()
    optimizer = optim.Adam(model.parameters(), lr=0.05)
    # 训练模型
    for epoch in range(num_epochs):
        model.train()
        optimizer.zero_grad()
        # 前向传播
        outputs = model(train_data)
        loss = criterion(outputs, target_data)
        # 反向传播和优化
        loss.backward()
        optimizer.step()
        if (epoch + 1) % 10 == 0:
            print(f'Epoch [{epoch + 1}/{num_epochs}], Loss: {loss.item():.4f}')
        if epoch == int(num_epochs/2):
            optimizer = optim.Adam(model.parameters(), lr=0.01)
    return model

def start_predicted(model, predicted_data):
    model.eval()
    with torch.no_grad():
        test_input = predicted_data.unsqueeze(0)  # 使用最后seq_length个时间步作为输入
        predicted = model(test_input)
        # print("Predicted:", predicted)
    return predicted

def start_all_train(hidden_size, num_layers, num_epochs, seq_length):
    r_data, b_data, r_train, r_target, b_train, b_target, r_mean, r_std, b_mean, b_std = trans_process_data(seq_length)
    # print(r_mean, r_std)
    r_size = 5
    r_model = start_train(r_size, hidden_size, r_size, num_layers, r_train, r_target, num_epochs)
    predicted_data = r_data[-seq_length:]
    r_predicted = start_predicted(r_model, predicted_data)
    print("--------------------------bbbbb-------------------------------------------")
    b_size = 2
    b_model = start_train(b_size, hidden_size, b_size, num_layers, b_train, b_target, num_epochs)
    predicted_data = b_data[-seq_length:]
    b_predicted = start_predicted(b_model, predicted_data)

    print(r_predicted)
    print(b_predicted)

    r_predicted = r_predicted * r_std + r_mean
    b_predicted = b_predicted * b_std + b_mean

    print(r_predicted)
    print(b_predicted)

    return r_predicted, b_predicted

if __name__ == '__main__':
    hidden_size = 20
    num_layers = 3
    num_epochs = 1000
    seq_length = 10

    r_predicted, b_predicted = start_all_train(hidden_size, num_layers, num_epochs, seq_length)
    # print(r_predicted)
    # print(b_predicted)

2.随机预测

下面是随机选号预测

import random
import numpy as np

from collections import Counter

# 大乐透和双色球不一样
r_len = 5
r_num = 35

b_len = 2
b_num = 12

# 双色球
# r_len = 6
# r_num = 33
#
# b_len = 1
# b_num = 16

number = 100000000
li = []

li_r = []
li_b = []
for i in range(number):

    r_li = random.sample(range(1, r_num+1), r_len)
    b_li = random.sample(range(1, b_num+1), b_len)
    li_r.extend(r_li)
    li_b.extend(b_li)
    print(i)

counter_li_r = Counter(li_r)
counter_li_b = Counter(li_b)

most_common_li_r = counter_li_r.most_common(r_len)
most_common_li_b = counter_li_b.most_common(b_len)
most_common_li_r = list(map(lambda x: x[0], most_common_li_r))
most_common_li_b = list(map(lambda x: x[0], most_common_li_b))
most_common_li_r.sort()
most_common_li_b.sort()

li = most_common_li_r
li.extend(most_common_li_b)
print("most: ", li)

most_least_li_r = counter_li_r.most_common()[-r_len-1:-1]
most_least_li_b = counter_li_b.most_common()[-b_len-1:-1]
most_least_li_r = list(map(lambda x: x[0], most_least_li_r))
most_least_li_b = list(map(lambda x: x[0], most_least_li_b))
most_least_li_r.sort()
most_least_li_b.sort()
li = most_least_li_r
li.extend(most_least_li_b)
print("least: ", li)

好运来，恭喜中一等奖