基于粒子群优化LSTM的股票预测模型

宋刚; 张云峰; 包芳勋; 秦超

doi:10.13700/j.bh.1001-5965.2019.0388

基于粒子群优化LSTM的股票预测模型

doi: 10.13700/j.bh.1001-5965.2019.0388

宋刚^{1, 2},
张云峰^{1, 2, ,},
包芳勋³,
秦超^{1, 2}

1.
山东财经大学计算机科学与技术学院, 济南 250014
2.
山东财经大学山东省数字媒体技术重点实验室, 济南 250014
3.
山东大学数学学院, 济南 250100

基金项目:

国家自然科学基金 61672018

国家自然科学基金 61772309

国家自然科学基金-浙江两化融合联合基金 U1609218

山东省重点研发计划 2016GSF120013

山东省重点研发计划 2017GGX10109

山东省重点研发计划 2018GGX101013

山东省高等学校优势学科人才队伍培育计划

山东省自然科学杰出青年 ZR2018JL022

山东省自然科学基金 ZR2019MF051

详细信息

作者简介:
宋刚男, 硕士研究生。主要研究方向:机器学习、数据分析

张云峰男, 博士, 教授, 博士生导师。主要研究方向:计算机视觉、数据分析

通讯作者:
张云峰, E-mail: yfzhang@sdufe.edu.cn

中图分类号: TP183
计量
- 文章访问数: 2463
- HTML全文浏览量: 415
- PDF下载量: 1064
- 被引次数: 0
出版历程
- 收稿日期: 2019-07-10
- 录用日期: 2019-08-23
- 网络出版日期: 2019-12-20

Stock prediction model based on particle swarm optimization LSTM

SONG Gang^{1, 2},
ZHANG Yunfeng^{1, 2
, ,},
BAO Fangxun³,
QIN Chao^{1, 2}

1.
School of Computer Science and Technology, Shandong University of Finance and Economics, Jinan 250014, China
2.
Shandong Key Laboratory of Digital Media Technology, Shandong University of Finance and Economics, Jinan 250014, China
3.
School of Mathematics, Shandong University, Jinan 250100, China

Funds:

National Natural Science Foundation of China 61672018

National Natural Science Foundation of China 61772309

National Natural Science Foundation of China-Zhejiang Two Integration Joint Fund U1609218

Key Research and Development Project of Shandong Province 2016GSF120013

Key Research and Development Project of Shandong Province 2017GGX10109

Key Research and Development Project of Shandong Province 2018GGX101013

Fostering Project of Dominant Discipline and Talent Team of Shandong Province Higher Education Institutions

Natural Science Foundation for Excellent Youth of Shandong Province ZR2018JL022

Natural Science Foundation of Shandong Province ZR2019MF051

More Information

Corresponding author: ZHANG Yunfeng, E-mail: yfzhang@sdufe.edu.cn

摘要

摘要:
为了提高股票时间序列预测精度，增强预测模型结构参数可解释性，提出一种基于自适应粒子群优化（PSO）的长短期记忆（LSTM）股票价格预测模型（PSO-LSTM），该模型在LSTM模型的基础上进行改进和优化，因此擅长处理具有长期依赖关系的、复杂的非线性问题。通过自适应学习策略的PSO算法对LSTM模型的关键参数进行寻优，使股票数据特征与网络拓扑结构相匹配，提高股票价格预测精度。实验分别以沪市、深市、港股股票数据构建了PSO-LSTM模型，并对该模型的预测结果与其他预测模型进行比较分析。结果表明，基于自适应PSO的LSTM股票价格预测模型不但提高了预测准确度，而且具有普遍适用性。
- 粒子群优化(PSO) /
- LSTM神经网络 /
- 自适应 /
- 股票价格预测 /
- 预测精度
Abstract:
This paper proposes a stock price prediction model based on particle swarm optimization long short-term memory (PSO-LSTM). This model improves and optimizes the LSTM model, which makes it more appropriate for analyzing relationships such as long-term dependency and for solving complex nonlinear problems. Through finding the key parameters in LSTM model by the PSO algorithm with adaptive learning strategy, the stock data feature matches the network topology structure, and the model's prediction accuracy of stock price is improved. In the experiment, PSO-LSTM models are constructed respectively based on the stock datasets from Shanghai, Shenzhen and Hong Kong, and then they are compared to other prediction models. The comparison results show that the PSO-LSTM stock price prediction model achieves higher prediction accuracy and has general applicability.
- particle swarm optimization (PSO) /
- LSTM neural network /
- adaptive /
- stock price forecast /
- prediction accuracy

HTML全文

图 1 LSTM单元结构

Figure 1. LSTM unit structure

下载: 全尺寸图片幻灯片

图 2 PSO-LSTM模型架构

Figure 2. PSO-LSTM model architecture

下载: 全尺寸图片幻灯片

图 3 上海浦发银行各模型预测结果比较

Figure 3. Comparison of prediction results of various models for Shanghai Pudong Development Bank

下载: 全尺寸图片幻灯片

图 4 LSTM与PSO-LSTM预测结果比较

Figure 4. Comparison of prediction results between LSTM and PSO-LSTM

下载: 全尺寸图片幻灯片

图 5 五粮液各模型预测结果比较

Figure 5. Comparison of prediction results of various models for Wuliangye

下载: 全尺寸图片幻灯片

图 6 恒隆集团各模型预测结果比较

Figure 6. Comparison of prediction results of various models for Hang Lung Group

下载: 全尺寸图片幻灯片

表 1 上海浦发银行各模型评价指标比较

Table 1. Comparison of various models' evaluation indicators for Shanghai Pudong Development Bank

模型	RMSE	MAPE/%	MAE	MSE	R²
ARIMA	0.1609	3.6176	0.3809	0.0259	0.9833
SVM	0.1847	1.4149	0.1482	0.0341	0.9666
MLP	0.1924	1.4665	0.1551	0.0374	0.9832
RNN	0.2082	1.5145	0.1598	0.0433	1.3423
LSTM	0.1799	1.3792	0.1451	0.0323	1.1518
PSO-LSTM	0.1420	1.0369	0.1068	0.0202	1.0120

下载: 导出CSV

表 2 LSTM与PSO-LSTM评价指标比较

Table 2. Comparison of evaluation indicators between LSTM and PSO-LSTM

模型	RMSE
模型	第1组	第2组	第3组	第4组	第5组
LSTM	0.1072	0.1061	0.1442	0.1629	1.1534
PSO-LSTM	0.1174	0.0913	0.1361	0.1240	0.1149

下载: 导出CSV

表 3 五粮液各模型评价指标比较

Table 3. Comparison of various models' evaluation indicators for Wuliangye

模型	RMSE	MAPE/%	MAE	MSE	R²
ARIMA	1.7814	11.1753	6.2734	3.2851	1.0178
SVM	1.6896	2.4872	1.3764	2.8548	0.7726
MLP	1.8892	2.6822	1.3879	3.4882	1.0605
RNN	1.6549	2.3478	1.3247	2.7203	1.0344
LSTM	1.6013	2.1247	1.1918	2.5734	1.0571
PSO-LSTM	1.3427	1.7808	1.0342	1.8028	1.0124

下载: 导出CSV

表 4 恒隆集团各模型评价指标比较

Table 4. Comparison of various models' evaluation indicators for Hang Lung Group

模型	RMSE	MAPE/%	MAE	MSE	R²
ARIMA	0.3582	4.3628	0.8921	0.1282	1.2391
SVM	0.5216	2.1204	0.4247	0.2721	0.5797
MLP	0.3439	1.2296	0.2529	0.1183	0.9695
RNN	0.3844	1.5902	0.3242	0.1478	0.9568
LSTM	0.3649	1.4834	0.3026	0.1332	0.9567
PSO-LSTM	0.2845	1.0472	0.2366	0.0812	1.0041

下载: 导出CSV

参考文献(19)

[1]	SI Y W, YIN J.OBST-based segmentation approach to financial time series[J]. Engineering Applications of Artificial Intelligence, 2013, 26(10):2581-2596. doi: 10.1016/j.engappai.2013.08.015
[2]	COSTA-JUSSÀ, MARTA R, ALLAUZEN A, et al.Introduction to the special issue on deep learning approaches for machine translation[J]. Computer Speech & Language, 2017, 46(11):367-373. http://www.wanfangdata.com.cn/details/detail.do?_type=perio&id=fda55bcc89ecd0a7b03110f018df42b7
[3]	FAYEK H M, LECH M, CAVEDON L.Evaluating deep learning architectures for speech emotion recognition[J]. Neural Network, 2017, 92(2):60-68. http://www.wanfangdata.com.cn/details/detail.do?_type=perio&id=8888b33e167101c6f2565c44aafcf9b1
[4]	XING J L, LI K, HU W M, et al.Diagnosing deep learning mod- els for high accuracy age estimation from a single image[J]. Pattern Recognition 2017, 66(1):106-116.
[5]	HOCHREITER S, SCHMIDHUBER J.Long short-term memory[J]. Neural Computation, 1997, 9(8):1735-1780. doi: 10.1162/neco.1997.9.8.1735
[6]	DEIHIMI A, ORANG O, SHOWKATI H.Short-term electric load and temperature forecasting using wavelet echo state networks with neural reconstruction[J]. Energy, 2013, 57(3):382-401. http://www.wanfangdata.com.cn/details/detail.do?_type=perio&id=3dc28149fa954940d688884f9c8a59eb
[7]	CAI M, LIU J.Maxout neurons for deep convolutional and LSTM neural networks in speech recognition[J]. Speech Communication, 2016, 77(3):53-64. http://www.wanfangdata.com.cn/details/detail.do?_type=perio&id=400b43137b9aaf4b333d27dbeae6100f
[8]	ZHOU C, SUN C, LIU Z, et al.A C-LSTM neural network for text classification[J]. Computer Science, 2015, 1(4):39-44.
[9]	CHEN K, ZHOU Y, DAI F.A LSTM-based method for stock returns prediction: A case study of China stock market[C]//IEEE International Conference on Big Data.Piscataway, NJ: IEEE Press, 2015: 2823-2824.
[10]	JIA H J.Investigation into the effectiveness of long short term memory networks for stock price prediction[EB/OL]. (2016-03-25)[2019-07-10]. https://arxu-org/abs/1603.07893v1.
[11]	NELSON D M Q, PEREIRA A C M, DEOLIVEIRA R A, et al.Stock market's price movement prediction with LSTM neural networks[C]//International Joint Conference on Neural Networks.Piscataway, NJ: IEEE Press, 2017: 1419-1426.
[12]	FISCHER T, KRAUSS C.Deep learning with long short-term memory networks for financial market predictions[J]. European Journal of Operational Research, 2018, 270(2):654-669. doi: 10.1016/j.ejor.2017.11.054
[13]	RODRIGUEZ A, LAIO A.Clustering by fast search and find of density peaks[J]. Science, 2014, 344(6191):1492-1496. doi: 10.1126/science.1242072
[14]	胡旺, 李志蜀.一种更简化而高效的粒子群优化算法[J].软件学报, 2007, 18(4):861-868. http://d.old.wanfangdata.com.cn/Periodical/rjxb200704008 HU W, LI Z S.A simpler and more effective particle swarm optimization algorithm[J]. Journal of Software, 2007, 18(4):861-868(in Chinese). http://d.old.wanfangdata.com.cn/Periodical/rjxb200704008
[15]	ADEBIYI A A, ADEWUMI A O, AYO C K.Stock price prediction using the ARIMA model[C]//UKSIM-AMSS International Conference on Computer Modelling & Simulation.Piscataway, NJ: IEEE Press, 2015: 106-112.
[16]	LIN Y, GUO H, HU J.An SVM-based approach for stock market trend prediction[C]//International Joint Conference on Neural Networks.Piscataway, NJ: IEEE Press, 2013: 1-7.
[17]	USMANI M, ADIL S H, RAZA K, et al.Stock market prediction using machine learning techniques[C]//International Conference on Computer & Information Sciences.Piscataway, NJ: IEEE Press, 2016: 322-327.
[18]	RATHER A M, AGARWAL A, SASTRY V.Recurrent neural network and a hybrid model for prediction of stock returns[J]. Expert Systems with Applications, 2015, 42(6):3234-3241. doi: 10.1016/j.eswa.2014.12.003
[19]	MCNALLY S, ROCHE J, CATON S.Predicting the price of Bitcoin using Machine Learning[C]//26th Euromicro International Conference on Parallel Distributed and Network-based Processing.Piscataway, NJ: IEEE Press, 2018: 339-343.