logo
科言猫学术研究的AI总结
language
手机端手机端
  • Phone Feature微信小程序
简体中文简体中文
notice消息通知
博客博客
delete
arrow
返回

Modelradar: aspect-based forecast evaluation

delete2025-09-09
delete0
delete
OA
AI
V
Vítor Cerqueira *
L
Luís Roque
C
Carlos Soares
DOI：10.1007/s10994-025-06877-zdelete
delete原文链接
delete分享
delete收藏
查看原文
摘要

摘要

En 中文
准确评估预测模型对于确保可靠的预测至关重要。当前评估和比较预测模型的实践侧重于使用SMAPE等指标将性能汇总为单一分数。尽管方便，但将所有样本的性能进行平均会淡化模型在不同条件下行为的相关信息。这一局限性在时间序列预测中尤其成问题，因为在数据集中对时间步长、预测范围和多个时间序列进行多层平均可能会掩盖相关的性能变化。我们通过提出ModelRadar框架来应对这一局限，该框架用于从多个方面评估单变量时间序列预测模型，例如平稳性、异常存在或预测范围。我们通过比较24种预测方法（包括经典方法和不同的机器学习算法）来证明该框架的优势。PatchTST，一种最先进的基于transformer的神经网络架构，总体表现最佳，但其优势因预测条件而异。例如，就预测范围而言，我们发现PatchTST（以及其他神经网络）仅在多步提前预测中优于经典方法。另一个相关见解是，经典方法（如ETS或Theta）在异常存在的情况下表现出显著更强的鲁棒性。这些及其他发现突显了面向方面的模型评估对实践者和研究人员的重要性。ModelRadar可作为Python软件包使用。
Keyword:
Time Series
Forecasting
Evaluation
AI总结

AI总结

对已上传原文的论文进行重点信息的提取，主要内容包括：简要概述、研究摘要、背景介绍、关键亮点、图文解析、展望与总结。

期刊

Machine Learning 封面图
Machine Learning
IF:
2.9
论文数:
2.7K
被引数:
3.4W

机构

F
faculdade de engenharia
学者数:
41
论文数: 18
被引数: 0
引用论文

引用论文

TESTING THE NULL HYPOTHESIS OF STATIONARITY AGAINST THE ALTERNATIVE OF A UNIT-ROOT - HOW SURE ARE WE THAT ECONOMIC TIME-SERIES HAVE A UNIT-ROOT
errJOURNAL OF ECONOMETRICS
IF4
err1992-10-01
err7.5K
PREAI
errKWIATKOWSKI, D; PHILLIPS, PCB; SCHMIDT, P; SHIN, YC
err分享
err收藏
Neural Network Models for Time Series Forecasts
errManagement Science
IF0
err1996-07-01
err0
PREAI
errTim Hill; Marcus O'Connor; William Remus
err分享
err收藏
Forecast evaluation for data scientists: common pitfalls and best practices
errData Mining and Knowledge Discovery
IF0
err2022-12-02
err0
errOAAI
errHansika Hewamalage; Klaus Ackermann; Christoph Bergmeir
err分享
err收藏
On the selection of forecasting accuracy measures
errJournal of the Operational Research Society
IF0
err2021-04-13
err0
errOAAI
errDiamantis Koutsandreas; Evangelos Spiliotis; Fotios Petropoulos; Vassilios Assimakopoulos
err分享
err收藏
Automatic time series forecasting: The forecast package for R自动时间序列预测: R的预测包
errJOURNAL OF STATISTICAL SOFTWARE
IF8.1
err2008-01-01
err2.5K
errOAAI
errHyndman, Rob J.; Khandakar, Yeasmin
err分享
err收藏
The tourism forecasting competition
errINTERNATIONAL JOURNAL OF FORECASTING
IF7.1
err2011-07-01
err199
errOAAI
errAthanasopoulos, George; Hyndman, Rob J.; Song, Haiyan; Wu, Doris C.
err分享
err收藏
Criteria for classifying forecasting methods
errINTERNATIONAL JOURNAL OF FORECASTING
IF7.1
err2020-01-01
err90
errOAAI
errJanuschowski, Tim; Gasthaus, Jan; Wang, Yuyang; Salinas, David; Flunkert, Valentin; Bohlke-Schneider, Michael; Callot, Laurent
err分享
err收藏
Temporal Fusion Transformers for interpretable multi-horizon time series forecasting用于可解释的多水平时间序列预测的时间融合变换器
errINTERNATIONAL JOURNAL OF FORECASTING
IF7.1
err2021-10-01
err670
errOAAI
errLim, Bryan; Arik, Sercan O.; Loeff, Nicolas; Pfister, Tomas
err分享
err收藏
An Empirical Comparison of Machine Learning Models for Time Series Forecasting
errEconometric Reviews
IF0
err2010-08-30
err0
PREAI
errNesreen K. Ahmed; Amir F. Atiya; Neamat El Gayar; Hisham El-Shishiny
err分享
err收藏
DeepAR: Probabilistic forecasting with autoregressive recurrent networksDeepAR: 自回归递归网络的概率预测
errINTERNATIONAL JOURNAL OF FORECASTING
IF7.1
err2020-07-01
err1.0K
errOAAI
errSalinas, David; Flunkert, Valentin; Gasthaus, Jan; Januschowski, Tim
err分享
err收藏
学者 查看更多内容