Code and data for SF-VMD-LSTM model and other decomposition ensemble models

This data repository contains code and data for research article, namely, “Decomposition ensemble model based on variational mode decomposition and long short-term memory for streamflow forecasting”, which will be (or have already been) published in the Journal of Hydrology. The streamflow data sets (daily streamflow series (01/01/1967-31/12/2014) of Yangxian station (yx), Han River and Zhangjiashan station (zjs), Jing River, China) used to build the proposed model are in the “time_series” directory. The fundamental code for decomposing streamflow data, deciding input predictors and output target, generating machine learning samples, building long short-term memory (LSTM) models and evaluating the model performance are organized in “tools” directory. The execution code for forecasting different streamflow series (zjs and yx) using different decomposition algorithms (e.g., variational mode decomposition (VMD), ensemble empirical mode decomposition (EEMD), discrete wavelet transform (DWT) or non-decomposition-based (orig)) are organized in “projects” directory (e.g., “zjs_vmd/projects/”). To reproduce the results of this paper, follow the instructions given in “readme.md”. Note that the same results demonstrated in this paper cannot be reproduced but similar results should be reproduced.本数据集仓库包含研究论文《基于变分模态分解与长短期记忆网络的径流预报分解集成模型》（Decomposition ensemble model based on variational mode decomposition and long short-term memory for streamflow forecasting）的配套代码与数据，该论文已或将刊发于《水文学杂志》（Journal of Hydrology）。 用于构建所提模型的径流数据集，涵盖中国汉江洋县站（yx）与泾河张家山站（zjs）1967年1月1日至2014年12月31日的日径流序列，相关数据存储于"time_series"目录中。 实现径流数据分解、确定输入预测因子与输出目标、生成机器学习样本、构建长短期记忆（long short-term memory, LSTM）模型以及评估模型性能的基础代码，收纳于"tools"目录下。 使用不同分解算法（如变分模态分解（variational mode decomposition, VMD）、集合经验模态分解（ensemble empirical mode decomposition, EEMD）、离散小波变换（discrete wavelet transform, DWT）以及无分解模式（non-decomposition-based, orig））对不同径流序列（zjs与yx）开展预报的执行代码，存储于"projects"目录下（例如"zjs_vmd/projects/"路径）。 若需复现本论文的实验结果，请遵循"readme.md"中给出的操作说明。需注意：本论文展示的完全一致结果无法复现，但可复现相近的实验结果。