| 摘 要: 针对金融时间序列数据的高噪声、时间依赖性等问题,提出了一种人工蜂群算法-长短期记忆-门控单元(ABC-LSTM-GRU)混合模型。该模型综合利用长短期记忆网络(LSTM)和门控循环单元(GRU)循环神经网络,更全面地捕捉时间序列中的长期和短期关系。在特征处理阶段,通过相关性分析对特征进行筛选,同时采用奇异谱分析(SSA)对数据进行分解,得到高频、中频和低频三个部分。在模型的超参数优化中,采用了改进后的人工蜂群算法(ABC),以提高模型的性能。为验证ABC-LSTM-GRU混合模型的有效性,选择NIFTY-50股票指数进行实证分析。实验结果对比显示,ABC-LSTM-GRU混合模型在时间序列预测方面的表现更佳,相较于LSTM与GRU模型,其在均方根误差(RMSE)指标上分别降低了28.3%与21.5%,显示出更为准确的预测性能。 |
| 关键词: GRU;LSTM;ABC;SSA;股市预测 |
|
中图分类号: TP391
文献标识码: A
|
|
| Time Series Decomposition and Prediction Model Based on ABC-LSTM-GRU |
|
ZHU Zijing, HE Liwen
|
(School of Internet of Things, Nanjing University of Posts and Telecommunications, Nanjing 210003, China)
zhon0713@qq.com; helw@njupt.edu.cn
|
| Abstract: This paper proposes a hybrid model of Artificial Bee Colony-Long Short Term Memory-Gated Recurrent Unit (ABC-LSTM-GRU) to address the issues of high noise and time dependence of financial time series data. This model integrates LSTM and GRU networks to comprehensively capture the long and short term relationships in time series. In the feature processing stage, features are screened by correlation analysis, and the data is decomposed by Singular Spectrum Analysis (SSA) into three parts, high frequency, medium frequency and low frequency. In the hyperparameter optimization of the model, an improved Artificial Bee Colony (ABC) algorithm is used to improve the performance of the model. In order to verify the effectiveness of ABC-LSTM-GRU model, NIFTY-50 stock index is selected for empirical analysis. The comparison of experimental results shows that ABC-LSTM-GRU hybrid model has better performance in time series prediction, and compared with LSTM and GRU model, it reduces the Root Mean Square Error (RMSE) index by 28.3% and 21.5% , respectively, demonstrating more excellent prediction performance. |
| Keywords: GRU; LSTM; ABC; SSA; stock market forecast |
