Big Data Mining and Analytics  2021, Vol. 4 Issue (4): 266-278    DOI: 10.26599/BDMA.2021.9020011
A Deep-Learning Prediction Model for Imbalanced Time Series Data Forecasting
Chenyu Hou(),Jiawei Wu(),Bin Cao(),Jing Fan*()
College of Computer Science and Technology, Zhejiang University of Technology, Hangzhou 310023, China

Abstract

Time series forecasting has attracted wide attention in recent decades. However, some time series are imbalanced and show different patterns between special and normal periods, leading to the prediction accuracy degradation of special periods. In this paper, we aim to develop a unified model to alleviate the imbalance and thus improving the prediction accuracy for special periods. This task is challenging because of two reasons: (1) the temporal dependency of series, and (2) the tradeoff between mining similar patterns and distinguishing different distributions between different periods. To tackle these issues, we propose a self-attention-based time-varying prediction model with a two-stage training strategy. First, we use an encoder-?decoder module with the multi-head self-attention mechanism to extract common patterns of time series. Then, we propose a time-varying optimization module to optimize the results of special periods and eliminate the imbalance. Moreover, we propose reverse distance attention in place of traditional dot attention to highlight the importance of similar historical values to forecast results. Finally, extensive experiments show that our model performs better than other baselines in terms of mean absolute error and mean absolute percentage error.

Received: 21 May 2021      Published: 30 August 2021
Fund:  National Key R&D Program of China(2018YFB1402800);Fundamental Research Funds for the Provincial Universities of Zhejiang(RF-A2020007)
Corresponding Authors: Jing Fan     E-mail: houcy@zjut.edu.cn;wujw@zjut.edu.cn;bincao@zjut.edu.cn;fanjing@zjut.edu.cn
About author: Chenyu Hou received the BEng degree from Zhejiang University of Technology, Hangzhou, China in 2016. He is now a PhD student at Zhejiang University of Technology. He has published many papers on international journals and conferences, like TKDE, CIKM, and WWWJ, etc. His research interests are database and data mining.|Jiawei Wu received the BEng degree from Zhejiang University of Technology, Hangzhou, China in 2017. He is now a PhD student at Zhejiang University of Technology. He has published several papers on international journals and conferences, like TOIT, MONET, etc. His research interest is natural language processing.|Bin Cao received the PhD degree in computer science from Zhejiang University, China in 2013. He then worked as a research associate at Hongkong University of Science and Technology and Noah’s Ark Lab, Huawei. He joined Zhejiang University of Technology, Hangzhou, China in 2014, and is now an associate professor at the College of Computer Science and Technology, Zhejiang University of Technology. He has published more than 30 papers on many international authoritative journals and conferences, including TKDE, TSC, and CIKM. His research interests include data mining and natural language processing.|Jing Fan received the BEng, MEng, and PhD degrees in computer science from Zhejiang University, China in 1990, 1993 and 2003, respectively. She is now a professor at the College of Computer Science and Technology, Zhejiang University of Technology, China. She is a director of China Computer Federation (CCF). She has published more than 100 papers on many international journals and conferences, including TSC, TKDE, and IEEE Virtual Reality. Her current research interests include service computing, virtual reality, and intelligent interaction.