层次时间序列预测方法与应用综述

doi:10.11896/jsjkx.241000139

摘要/Abstract

摘要： 层次时间序列是解决具有层级约束的多元时间序列,上层节点的数据是其所有子节点数据的累加。层次时间序列预测的主要难点是在准确预测每个序列的同时,还要保证不同级别之间的一致性,即预测在层次结构中满足累加的约束。随着大规模数据的涌现,这一复杂而具有挑战性的问题展现出更大的研究价值和广泛的应用前景。通过对层次时间序列预测相关方法和文献的综述,从分类方法和应用理论两个方面进行总结和归纳,同时探讨了该技术面临的挑战和实际应用中存在的缺口。分析表明,层次时间序列预测方法主要可分为预测模型和修订模型两个阶段,逐步引入机器学习和深度学习方法,并演化为将预测和修订模型融合的端对端方法。这些方法广泛应用于商业运营和政府治理领域。在未来的研究趋势方面,首先需要关注海量数据对两阶段方法预测准确度的影响;其次是深入研究端对端层次时间序列预测模型,以避免两阶段参数不连动的问题。此外,政府管理和商业运营的研究可以侧重于对具体问题导致不同层级关注度差异进行建模。

关键词: 层次时间序列预测, 一致性, 机器学习, 深度学习

Abstract: Hierarchical time series involve multiple time series with hierarchical constraints,where the data of the upper-level nodes are the cumulative sum of all their child node data.The main challenge in hierarchical time series forecasting is to accurately predict each series while ensuring consistency across different levels,i.e.,the forecast must satisfy the additive constraints within the hierarchical structure.With the emergence of large-scale data,this complex and challenging problem has demonstrated greater research value and a broad range of application prospects.This study reviews the literature on hierarchical time series forecasting methods,summarizing and generalizing from the aspects of classification methods and theoretical applications,while also discussing the challenges faced by this technology and the gaps in practical applications.The analysis indicates that hierarchical time series forecasting methods can mainly be divided into two stages,which are forecasting models and revision models,gradually introducing machine learning and deep learning methods,and evolving into end-to-end methods that integrate forecasting and revision models.These methods are widely applied in the fields of business operations and government governance.In terms of future research trends,the first area of focus should be the impact of massive data on the accuracy of two-stage method forecasts,followed by in-depth research on end-to-end hierarchical time series forecasting models to avoid the issue of non-coherent parameters between stages.Additionally,research in government management and business operations can focus on modeling the differences in attention levels at various hierarchical levels caused by specific issues.

Key words: Hierarchical time series forecasting, Consistency, Machine learning, Deep learning

中图分类号:

TP181

向易, 丛丽丽, 王玮鹏, 周晓航. 层次时间序列预测方法与应用综述[J]. 计算机科学, 2025, 52(11A): 241000139-7. https://doi.org/10.11896/jsjkx.241000139

XIANG Yi, CONG Lili, WANG Weipeng, ZHOU Xiaohang. Comprehensive Review of Hierarchical Time Series Forecasting Methods and Applications[J]. Computer Science, 2025, 52(11A): 241000139-7. https://doi.org/10.11896/jsjkx.241000139

参考文献

[1]PETROPOULOS F,APILETTI D,ASSIMAKOPOULOS V,et al. Forecasting:Theory and practice[J].International Journal of Forecasting,2022,38(3):705-871.
[2]ZHOU F,PAN C,MA L,et al.SLOTH:Structured learning and task-based optimization for time series forecasting on hierarchies[C]//Proceedings of the AAAI Conference on Artificial Intelligence.2023:11417-11425.
[3]TAIEB S B,TAYLOR J W,HYNDMAN R J.Coherent probabilistic forecasts for hierarchical time series[C]//International Conference on Machine Learning.2017:3348-3357.
[4]SEEGER M,RANGAPURAM S,WANG Y,et al.Approximate Bayesian inference in linear state space models for intermittent demand forecasting at scale[J].arXiv:1709.07638,2017.
[5]SEEGER M,SALINAS D,FLUNKERT V.Bayesian intermit-tent demand forecasting for large inventories[C]//Proceedings of the 30th International Conference on Neural Information Processing Systems.2016:4653-4661.
[6]JEON J,PANAGIOTELIS A,PETROPOULOS F.Probabilistic forecast reconciliation with applications to wind power and electric load[J].European Journal of Operational Research,2019,279(2):364-379.
[7]TAIEB S B,TAYLOR J W,HYNDMAN R J.Hierarchicalprobabilistic forecasting of electricity demand with smart meter data[J].Journal of the American Statistical Association,2021,116(533):27-43.
[8]JANUSCHOWSKI T,KOLASSA S.3.2 A classification of business forecasting problems[M]//Business Forecasting:Practical Problems and Solutions.Germany:Wiley,2016:171.
[9]CHOPRA S,MEINDL P.Supply chain management.Strategy,planning & operation[M].Springer,2007.
[10]CAPLICE C G,SHEFFI Y.ESD.260J/1.260J/15.770J Logistics Systems,Fall 2003[M/OL].Massachusetts Institute of Technology,2003.
[11]BOX G E,JENKINS G M,REINSEL G C,LJUNG G M.Time series analysis:Forecasting and control[M].John Wiley & Sons,2015.
[12]MONTGOMERY D C,JENNINGS C L,KULAHCI M.Intro-duction to time series analysis and forecasting[M].John Wiley &Sons,2015.
[13]MANCUSO P,PICCIALLI V,SUDOSO A M.A machine learning approach for forecasting hierarchical time series[J].Expert Systems with Applications,2021,182:115102.
[14]GRUNFELD Y,GRILICHES Z.Is aggregation necessarily bad?[J].The Review of Economics and Statistics,1960,42(1):1-13.
[15]FLIEDNER G.An investigation of aggregate variable time se-ries forecast strategies with specific subaggregate time series statistical correlation[J].Computers & Operations Research,1999,26(10-11):1133-1149.
[16]ORCUTT G H,WATTS H W,EDWARDS J B.Data aggregation and information loss[J].The American Economic Review,1968,58(4):773-787.
[17]EDWARDS J B,ORCUTT G H.Should aggregation prior to estimation be the rule?[J].The Review of Economics and Statistics,1969,51(4):409-420.
[18]SCHWARZKOPF A B,TERSINE R J,MORRIS J S.Top-down versus bottom-up forecasting strategies[J].The International Journal of Production Research,1988,26(11):1833-1843.
[19]HYNDMAN R J,AHMED R A,ATHANASOPOULOS G,et al.Optimal combination forecasts for hierarchical time series[J].Computational Statistics & Data Analysis,2011,55(9):2579-2589.
[20]WICKRAMASURIYA S L,ATHANASOPOULOS G,HYND-MAN R J.Optimal forecast reconciliation for hierarchical and grouped time series through trace minimization[J].Journal of the American Statistical Association,2019,114(526):804-819.
[21]TAIEB S B,KOO B.Regularized regression for hierarchicalforecasting without unbiasedness conditions[C]//Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining.2019:1337-1347.
[22]SHLIFER E,WOLFF R W.Aggregation and proration in forecasting[J].Management Science,1979,25(6):594-603.
[23]ABOLGHASEMI M,HYNDMAN R J,TARR G,et al.Machine learning applications in time series hierarchical forecasting[J].arXiv preprint arXiv:1912.00370,2019.
[24]SHIRATORI T,KOBAYASHI K,TAKANO Y.Prediction of hierarchical time series using structured regularization and its application to artificial neural networks[J].Plos One,2020,15(11):e0242099.
[25]BURBA D,CHEN T.A trainable reconciliation method for hierarchical time-series[J].arXiv preprint arXiv:2101.01329,2021.
[26]MISHCHENKO K,MONTGOMERY M,VAGGI F.A self-supervised approach to hierarchical forecasting with applications to groupwise synthetic controls[J].arXiv:1906.10586,2019.
[27]GLEASON J L.Forecasting hierarchical time series with a regularized embedding space[C]//MileTS'20:6th KDD Workshop on Mining and Learning from Time Series.2020:883-894.
[28]HAN X,DASGUPTA S,GHOSH J.Simultaneously reconciled quantile forecasting of hierarchically related time series[C]//International Conference on Artificial Intelligence and Statistics.2021:190-198.
[29]RANGAPURAM S S,WERNER L D,BENIDIS K,et al.End-to-end learning of coherent probabilistic forecasts for hierarchical time series[C]//International Conference on Machine Lear-ning.PMLR,2021:8832-8843.
[30]WANG S,ZHOU F,SUN Y,et al.End-to-end modeling of hierarchical time series using autoregressive transformer and conditional normalizing flow-based reconciliation[C]//2022 IEEE International Conference on Data Mining Workshops(ICDMW).2022:1087-1094.
[31]XIANG Y,SUN H,TU W.HDResNet:Hierarchical-decomposition residual network for hierarchical time series forecasting[C]//2023 International Joint Conference on Neural Networks (IJCNN).2023:1-8.
[32]ROSTAMI-TABAR B,BABAI M Z,DUCQ Y,et al.Non-stationary demand forecasting by cross-sectional aggregation[J].International Journal of Production Economics,2015,170:297-309.
[33]MIRCETIC D,ROSTAMI-TABAR B,NIKOLICIC S,et al.Forecasting hierarchical time series in supply chains:An empirical investigation[J].International Journal of Production Research,2022,60(8):2514-2533.
[34]VILLEGAS M A,PEDREGAL D J.Supply chain decision support systems based on a novel hierarchical forecasting approach[J].Decision Support Systems,2018,114:29-36.
[35]ABOLGHASEMI M,HYNDMAN R J,SPILIOTIS E,et al.Model selection in reconciling hierarchical time series[J].Machine Learning,2022,111:739-789.
[36]ABOLGHASEMI M,TARR G,BERGMEIR C.Machine learning applications in hierarchical time series forecasting:Investigating the impact of promotions[J].International Journal of Forecasting,2024,40(2):597-615.
[37]SPILIOTIS E,ABOLGHASEMI M,HYNDMAN R J,et al.Hierarchical forecast reconciliation with machine learning[J].Applied Soft Computing,2021,112:107756.
[38]KARMY J P,MALDONADO S.Hierarchical time series fore-casting via support vector regression in the European travel retail industry[J].Expert Systems with Applications,2019,137:59-73.
[39]BENABDALLAH BENARMAS R,BEGHDAD BEY K.A deep learning hierarchical approach to road traffic forecasting[J].Journal of Forecasting,2024,43(5):1294-1307.
[40]ATHANASOPOULOS G,GAMAKUMARA P,PANAGIO-TELIS A,et al.Hierarchical forecasting[M]//Macroeconomic Forecasting in the Era of Big Data.Advanced Studies in Theoretical and Applied Econometrics.2020,52:689-719.
[41]BISAGLIA L,DI FONZO T,GIROLIMETTO D.Fully reconciled GDP forecasts from income and expenditure sides[R].ar-Xiv:2004.03864,2020.
[42]REDELICO F O,PROTO A N,AUSLOOS M.Hierarchicalstructures in the Gross Domestic Product per capita fluctuation in Latin American countries[J].Physica A:Statistical Mechanics and its Applications,2009,388(17):3527-3535.
[43]WEISS C.Essays in hierarchical time series forecasting and forecast combination[D].University of Cambridge,2018.

Metrics

Viewed

Full text

Abstract

Cited

Shared

Discussed