基于情感分析的LSTM预测股票走势

 import pandas as pd

    
 from snownlp import SnowNLP
    
  
    
 # 加载数据
    
 data = pd.read_excel('浦发银行.xlsx')
    
  
    
 # 确保时间列存在且格式正确。由于时间列包括具体时间，因此需要匹配包括时间的格式
    
 data['update_time'] = pd.to_datetime(data['update_time'], format='%Y/%m/%d %H:%M', errors='coerce')
    
 # 确保时间和标题都不是空值
    
 data = data.dropna(subset=['update_time', 'title'])
    
  
    
 # 情感分析，计算情感得分
    
 data['sentiment_score'] = data['title'].apply(lambda x: SnowNLP(x).sentiments)
    
  
    
 # 根据情感得分定义情感类型
    
 threshold = 0.5
    
 data['sentiment_type'] = data['sentiment_score'].apply(lambda x: 'positive' if x >= threshold else 'negative')
    
  
    
 # 对每天的数据分组并计算positive和negative的平均值
    
 result = data.groupby(data['update_time'].dt.date).agg({
    
     'title': 'count',  # 计算每天的标题数量
    
     'sentiment_score': ['mean', 'count'],  # 计算每天的平均情感得分及其数量
    
     'sentiment_type': lambda x: (x == 'positive').mean()  # 计算正面情感的比例
    
 }).reset_index()
    
  
    
 # 重命名列便于理解
    
 result.columns = ['Date', 'Title Count', 'Average Sentiment Score', 'Sentiment Count', 'Positive Sentiment Ratio']
    
  
    
 # 创建新的列pos和neg
    
 result['Positive'] = result.apply(lambda x: x['Average Sentiment Score'] if x['Positive Sentiment Ratio'] >= 0.5 else None, axis=1)
    
 result['Negative'] = result.apply(lambda x: x['Average Sentiment Score'] if x['Positive Sentiment Ratio'] < 0.5 else None, axis=1)
    
  
    
 # 使用0.5填充空缺值（根据需要可以调整这个值）
    
 result['Positive'].fillna(0.5, inplace=True)
    
 result['Negative'].fillna(0.5, inplace=True)
    
  
    
 # 保存到Excel文件
    
 result.to_excel('分析结果.xlsx', index=False)
    
    
    
    
    python
    
    
![](https://ad.itadn.com/c/weblog/blog-img/images/2025-08-17/9Gk0T2eyM6vWDwEhXZ7bmOlxNazs.png)

 import pandas as pd

    
  
    
 # Load both Excel files
    
 result_path = '/mnt/data/分析结果.xlsx'
    
 data_result = pd.read_excel(result_path)
    
  
    
 bank_data_path = '/mnt/data/浦发银行.xlsx'
    
 data_bank = pd.read_excel(bank_data_path)
    
  
    
 # Ensure both date columns are in the same datetime format
    
 data_bank['日期'] = pd.to_datetime(data_bank['日期'], errors='coerce')
    
 data_result['Date'] = pd.to_datetime(data_result['Date'], errors='coerce')
    
  
    
 # Attempt to merge again
    
 merged_data_final = pd.merge(data_result, data_bank, left_on='Date', right_on='日期', how='inner')
    
  
    
 # Save the correctly merged data to a new Excel file
    
 final_merged_file_path = '/mnt/data/final_merged_data.xlsx'
    
 merged_data_final.to_excel(final_merged_file_path, index=False)
    
    
    
    
    python
    
    
![](https://ad.itadn.com/c/weblog/blog-img/images/2025-08-17/p8EmQrFc97DVSJPNbH5IRiA3yeft.png)

 #!/usr/bin/python3

    
 # -*- encoding: utf-8 -*-
    
 from this import s
    
  
    
 import matplotlib.pyplot as plt
    
 import numpy as np
    
 import tushare as ts
    
 import pandas as pd
    
 import torch
    
 from torch import nn
    
 import datetime
    
 import time
    
 from sklearn.metrics import r2_score  # Import r2_score from sklearn
    
  
    
 DAYS_FOR_TRAIN = 10
    
  
    
  
    
 class LSTM_Regression(nn.Module):
    
     """
    
     使用LSTM进行回归
    
   22.         参数：
    
     - input_size: feature size
    
     - hidden_size: number of hidden units
    
     - output_size: number of output
    
     - num_layers: layers of LSTM to stack
    
     """
    
  
    
     def __init__(self, input_size, hidden_size, output_size=1, num_layers=2):
    
     super().__init__()
    
  
    
     self.lstm = nn.LSTM(input_size, hidden_size, num_layers)
    
     self.fc = nn.Linear(hidden_size, output_size)
    
  
    
     def forward(self, _x):
    
     x, _ = self.lstm(_x)  # _x is input, size (seq_len, batch, input_size)
    
     s, b, h = x.shape  # x is output, size (seq_len, batch, hidden_size)
    
     x = x.view(s * b, h)
    
     x = self.fc(x)
    
     x = x.view(s, b, -1)  # 把形状改回来
    
     return x
    
  
    
  
    
 def create_dataset(data, days_for_train=5) -> (np.array, np.array):
    
     """
    
     根据给定的序列data，生成数据集
    
   48.         数据集分为输入和输出，每一个输入的长度为days_for_train，每一个输出的长度为1。
    
     也就是说用days_for_train天的数据，对应下一天的数据。
    
   51.         若给定序列的长度为d，将输出长度为(d-days_for_train+1)个输入/输出对
    
     """
    
     dataset_x, dataset_y = [], []
    
     for i in range(len(data) - days_for_train):
    
     _x = data[i:(i + days_for_train)]
    
     dataset_x.append(_x)
    
     dataset_y.append(data[:, 2][i + days_for_train])
    
     return (np.array(dataset_x), np.array(dataset_y))
    
  
    
  
    
 # 定义一个函数来对DataFrame的指定列进行归一化
    
 def normalize_columns(df, columns_to_normalize):
    
     for column in columns_to_normalize:
    
     df[column] = (df[column] - df[column].min()) / (df[column].max() - df[column].min())
    
     return df
    
  
    
  
    
 if __name__ == '__main__':
    
     t0 = time.time()
    
     data = pd.read_csv('浦发银行单一.csv', index_col='Date')  # 读取文件
    
     print(data.shape)
    
     data = data.astype('float32').values  # 转换数据类型
    
     data = data
    
     print(data.shape)
    
     print(data)
    
     for i in range(data.shape[1]):
    
     plt.plot(data[:, i])
    
     plt.savefig('data.png', format='png', dpi=200)
    
     plt.close()
    
     # 将价格标准化到0~1
    
     max_value = np.max(data[:, i])
    
     min_value = np.min(data[:, i])
    
     data[:, i] = (data[:, i] - min_value) / (max_value - min_value)
    
     print(data[:, i])
    
  
    
  
    
     dataset_x, dataset_y = create_dataset(data, DAYS_FOR_TRAIN)
    
  
    
     train_size = int(len(dataset_x) * 0.70)
    
  
    
     train_x = dataset_x[:train_size]
    
     train_y = dataset_y[:train_size]
    
  
    
    
    
     train_x = train_x.reshape(-1, 1, DAYS_FOR_TRAIN * 6)
    
     train_y = train_y.reshape(-1, 1, 1)
    
  
    
  
    
     train_x = torch.from_numpy(train_x)
    
     train_y = torch.from_numpy(train_y)
    
     #
    
     model = LSTM_Regression(6 * DAYS_FOR_TRAIN, 10, output_size=1, num_layers=4)  
    
  
    
     model_total = sum([param.nelement() for param in model.parameters()])  # 计算模型参数
    
     print("Number of model_total parameter: %.8fM" % (model_total / 1e6))
    
  
    
     train_loss = []
    
     loss_function = nn.MSELoss()
    
     optimizer = torch.optim.Adam(model.parameters(), lr=1e-2, betas=(0.9, 0.999), eps=1e-08, weight_decay=0)
    
     for i in range(350):
    
     out = model(train_x)
    
     loss = loss_function(out, train_y)
    
     loss.backward()
    
     optimizer.step()
    
     optimizer.zero_grad()
    
     train_loss.append(loss.item())
    
  
    
     # 将训练过程的损失值写入文档保存，并在终端打印出来
    
     with open('log.txt', 'a+') as f:
    
         f.write('{} - {}\n'.format(i + 1, loss.item()))
    
     if (i + 1) % 1 == 0:
    
         print('Epoch: {}, Loss:{:.5f}'.format(i + 1, loss.item()))
    
  
    
     # 画loss曲线
    
     plt.figure()
    
     plt.plot(train_loss, 'b', label='loss')
    
     plt.title("Train_Loss_Curve")
    
     plt.ylabel('train_loss')
    
     plt.xlabel('epoch_num')
    
     plt.savefig('loss.png', format='png', dpi=200)
    
     plt.close()
    
  
    
     # torch.save(model.state_dict(), 'model_params.pkl')  # 可以保存模型的参数供未来使用
    
     t1 = time.time()
    
     T = t1 - t0
    
     print('The training time took %.2f' % (T / 60) + ' mins.')
    
  
    
     tt0 = time.asctime(time.localtime(t0))
    
     tt1 = time.asctime(time.localtime(t1))
    
     print('The starting time was ', tt0)
    
     print('The finishing time was ', tt1)
    
  
    
     # for test
    
     model = model.eval()  # 转换成测试模式
    
     # model.load_state_dict(torch.load('model_params.pkl'))  # 读取参数
    
  
    
     # 注意这里用的是全集 模型的输出长度会比原数据少DAYS_FOR_TRAIN 填充使长度相等再作图
    
     dataset_x = dataset_x.reshape(-1, 1, DAYS_FOR_TRAIN * 6)  # (seq_size, batch_size, feature_size)
    
     dataset_x = torch.from_numpy(dataset_x)
    
  
    
     pred_test = model(dataset_x)  # 全量训练集
    
     # 的模型输出 (seq_size, batch_size, output_size)
    
     pred_test = pred_test.view(-1).data.numpy()
    
     pred_test = np.concatenate((np.zeros(DAYS_FOR_TRAIN), pred_test))  # 填充0 使长度相同
    
     assert len(pred_test) == len(data)
    
  
    
     # Calculate R^2 score
    
     r2 = r2_score(data[:, 3], pred_test)
    
     print(f'R^2 score: {r2:.4f}')
    
  
    
     plt.plot(pred_test, 'r', label='prediction')
    
     plt.plot(data[:, 3], 'b', label='real')
    
     plt.plot((train_size, train_size), (0, 1), 'g--')  # 分割线 左边是训练数据 右边是测试数据的输出
    
     plt.legend(loc='best')
    
     plt.title(f'Prediction vs Real (R^2 score: {r2:.4f})')
    
     plt.savefig('result.png', format='png', dpi=200)
    
     plt.show()
    
     plt.close()
    
  
    
    
    
    
    python
    
    
![](https://ad.itadn.com/c/weblog/blog-img/images/2025-08-17/psz8BlvSWt9iOZTDRNAXaMug3mnw.png)