Multivariate Time Series Forecasting
Multivariate time series forecasting aims to predict the future values of multiple interconnected time series, a crucial task across diverse fields like finance and healthcare. Current research emphasizes improving model accuracy and robustness, focusing on architectures like Transformers and Graph Neural Networks, as well as addressing challenges such as non-stationarity, missing data, and computational efficiency through techniques like frequency domain analysis and lightweight model designs. These advancements are significant for improving decision-making in various applications by providing more accurate and reliable predictions, particularly in scenarios with high-dimensional data and complex temporal dependencies.
Papers
BasisFormer: Attention-based Time Series Forecasting with Learnable and Interpretable Basis
Zelin Ni, Hang Yu, Shizhan Liu, Jianguo Li, Weiyao Lin
AutoMixer for Improved Multivariate Time-Series Forecasting on Business and IT Observability Data
Santosh Palaskar, Vijay Ekambaram, Arindam Jati, Neelamadhav Gantayat, Avirup Saha, Seema Nagar, Nam H. Nguyen, Pankaj Dayama, Renuka Sindhgatta, Prateeti Mohapatra, Harshit Kumar, Jayant Kalagnanam, Nandyala Hemachandra, Narayan Rangaraj
GCformer: An Efficient Framework for Accurate and Scalable Long-Term Multivariate Time Series Forecasting
YanJun Zhao, Ziqing Ma, Tian Zhou, Liang Sun, Mengni Ye, Yi Qian
TSMixer: Lightweight MLP-Mixer Model for Multivariate Time Series Forecasting
Vijay Ekambaram, Arindam Jati, Nam Nguyen, Phanwadee Sinthong, Jayant Kalagnanam