关键词:
弹性网络回归
股票价格预测
多重共线性
岭回归
Lasso回归
摘要:
本文基于弹性网络回归模型对股票价格进行了预测分析。通过收集国内某酒厂738个交易日的数据,选取开盘价、最高价、最低价、交易量、涨跌幅等作为自变量,以收盘价为因变量,分别应用线性回归、岭回归、Lasso回归及弹性网络回归四种模型进行预测分析。研究结果表明,弹性网络回归模型在处理多重共线性问题和变量筛选方面具有显著优势。通过交叉验证确定了最优的惩罚参数,使得模型的预测误差最小。最终,最高价、最低价、交易量和涨跌幅被筛选为影响股票收盘价的主要因素。本文的研究为股票价格预测提供了有效的方法和工具,并为金融市场的投资决策提供了重要参考。This study conducts a stock price prediction analysis based on the Elastic Net regression model. Using data from 738 trading days of a domestic brewery, the study selects opening price, highest price, lowest price, trading volume, and price change percentage as independent variables, with the closing price as the dependent variable. Four models are applied for prediction analysis: linear regression, ridge regression, Lasso regression, and Elastic Net regression. The results show that the Elastic Net regression model has significant advantages in handling multicollinearity issues and variable selection. The optimal penalty parameters are determined through cross-validation, minimizing the prediction error. Ultimately, the highest price, lowest price, trading volume, and price change percentage are identified as the key factors influencing stock closing prices. This research provides an effective method and tool for stock price prediction and offers important insights for financial market investment decisions.