Simultaneous model construction and noise reduction for hierarchical time series via Support Vector Regression

作者：

Highlights：

• A novel Support Vector Regression method is proposed for hierarchical time series.

• All the required regressors are constructed simultaneously in a single optimization problem.

• Our proposal pools information across levels of hierarchy.

• Best predictive performance is achieved in experiments on benchmark datasets.

摘要

•A novel Support Vector Regression method is proposed for hierarchical time series.•All the required regressors are constructed simultaneously in a single optimization problem.•Our proposal pools information across levels of hierarchy.•Best predictive performance is achieved in experiments on benchmark datasets.

论文关键词：Time series forecasting,Support vector machines,Support Vector Regression,Heterogeneity control,Hierarchical time series

论文评审过程：Received 28 October 2020, Revised 11 July 2021, Accepted 11 September 2021, Available online 14 September 2021, Version of Record 24 September 2021.

论文官网地址：https://doi.org/10.1016/j.knosys.2021.107492