[论文解读] Research on stock price forecast of general electric based on mixed CNN-LSTM model

Zeng‐Zhen Hu, Shen Bao+2 more|ArXiv.org|Jan 15, 2025

Energy Load and Power Forecasting被引用 9

一句话总结

论文构建了一个混合 CNN-LSTM 模型来预测 General Electric's stock price，使用 CNN 进行特征提取，LSTM 捕捉长期依赖，且使用带动态学习率和 L2 正则化的 SGD；它报告拟合良好但需要实时和极端市场测试。

ABSTRACT

Accurate stock price prediction is crucial for investors and financial institutions, yet the complexity of the stock market makes it highly challenging. This study aims to construct an effective model to enhance the prediction ability of General Electric's stock price trend. The CNN - LSTM model is adopted, combining the feature extraction ability of CNN with the long - term dependency handling ability of LSTM, and the Adam optimizer is used to adjust the parameters. In the data preparation stage, historical trading data of General Electric's stock is collected. After cleaning, handling missing values, and feature engineering, features with strong correlations to the closing price are selected and dimensionality reduction is performed. During model training, the data is divided into training, validation, and testing sets in a ratio of 7:2:1. The Stochastic Gradient Descent algorithm is used with a dynamic learning rate adjustment and L2 regularization, and the Mean Squared Error is used as the loss function, evaluated by variance, R - squared score, and maximum error. Experimental results show that the model loss decreases steadily, and the predicted values align well with the actual values, providing a powerful tool for investment decisions. However, the model's performance in real - time and extreme market conditions remains to be tested, and future improvements could consider incorporating more data sources.

研究动机与目标

• Motivate accurate stock price prediction for investors and financial institutions.
• Aim to improve GE stock price trend forecasting using a hybrid CNN-LSTM architecture.
• Apply data cleaning, feature engineering, and dimensionality reduction to enhance predictive signals.
• Evaluate model performance with standard regression metrics and prevent overfitting through regularization.

提出的方法

• Use CNN to extract short-term features from historical stock data.
• Combine CNN outputs with LSTM to capture long-term dependencies in price dynamics.
• Train with Stochastic Gradient Descent using a dynamic learning rate and L2 regularization.
• Optimize parameters with Adam in the abstract, and use MSE as the loss function.
• Split data into training, validation, and testing sets in a 7:2:1 ratio.
• Evaluate predictions using variance, R-squared, and maximum error.

实验结果

研究问题

• RQ1 Can a mixed CNN-LSTM model improve GE stock price forecasting compared to baseline methods?
• RQ2 What is the impact of feature engineering and dimensionality reduction on predictive performance?
• RQ3 How does the model perform under standard and near-term forecasting scenarios?
• RQ4 What are the limitations of real-time and extreme market condition predictions for this approach?

主要发现

• The model loss decreases steadily during training.
• Predicted values align well with actual GE stock prices.
• The approach provides a tool that could support investment decisions.
• Real-time performance and extreme market condition testing remain for future work.
• The study suggests incorporating more data sources could further improve results.