基于随机森林的A股股票涨跌预测研究

针对传统预测模型易陷入过拟合、缺失数据敏感、计算量大等不足,利用随机森林算法的双重随机性、处理数据集优异等特点,对A股股票涨跌预测进行研究。首先运用相关性分析对初始指标体系进行一次Spearman和二次Pearson筛选,去除指标体系中的冗余指标。然后对随机森林的各项重要参数进行优化,并对优化后的模型采用重要性估计方法以提升训练模型精确度。通过不同指标体系的对比,验证实验过程的正确性。最后,对比不同建模方法的实证预测结果,表明随机森林模型比传统机器学习方法二元logistic回归在性能上更优越,具备较高的预测准确度。

Forecast of A-Share Stock Change Based on Random Forest

In view of the shortcomings of traditional forecasting models such as easy overfitting, missing data sensitivity and large computation, a random forest algorithm was used to study the stock price change forecast, utilizing its benefits of double randomness excellent data processing performances. First, the correlation index analysis was carried out to select the initial index system once and twice. Next, the important parameters of the random forest were optimized, and then an importance estimation method was adopted to improve the accuracy of the training model. Through the comparison between different index systems, the correctness of the experimental process was verified. Finally, comparing the empirical results of different modeling methods, it is shown that the random forest model is superior to the binary logistic regression model and has higher prediction accuracy.