Bewertung der Verzerrung von Punktprognosen über mehrere Zeitreihen hinweg: Maßnahmen und visuelle Werkzeuge
Assessing point forecast bias across multiple time series: Measures and visual tools
Andrey Davydenko (andrey@live.co.uk) and
Paul Goodwin
Additional contact information
Andrey Davydenko: Independent Researcher
Paul Goodwin: University of Bath [Bath]
Post-Print from HAL
Abstract:
Measuring bias is important as it helps identify flaws in quantitative forecasting methods or judgmental forecasts. It can, therefore, potentially help improve forecasts. Despite this, bias tends to be under-represented in the literature: many studies focus solely on measuring accuracy. Methods for assessing bias in single series are relatively well-known and well-researched, but for datasets containing thousands of observations for multiple series, the methodology for measuring and reporting bias is less obvious. We compare alternative approaches against a number of criteria when rolling-origin point forecasts are available for different forecasting methods and for multiple horizons over multiple series. We focus on relatively simple, yet interpretable and easy-to-implement metrics and visualization tools that are likely to be applicable in practice. To study the statistical properties of alternative measures we use theoretical concepts and simulation experiments based on artificial data with predetermined features. We describe the difference between mean and median bias, describe the connection between metrics for accuracy and bias, provide suitable bias measures depending on the loss function used to optimise forecasts, and suggest which measures for accuracy should be used to accompany bias indicators. We propose several new measures and provide our recommendations on how to evaluate forecast bias across multiple series.
Keywords: forecasting; forecast bias; mean bias; median bias; MPE; AvgRel-metrics; AvgRelAME; AvgRelAMdE; RelAME; RelMdE; AvgRelME; AvgRelMdE; OPc; Mean Percentage Error; MAD/MEAN ratio; Overestimation Percentage corrected; OPc-diagram; OPc-boxplot; AvgRel-prefix; RelAMdE; RelME; absolute mean error; absolute median error; AvgRelRMSE; AvgRelMAE; AvgRelMSE; AvgRel-boxplots; statistical graphics; forecast evaluation workflow; FEW; FEW-L1; FEW-L2; pooled prediction-realization diagram; prediction-realization diagram; criteria for error measures; construct validity; target loss function; point forecast evaluation setup; PFES; forecasting competitions; testing for bias; geometric mean; optimal correction of forecasts; symmetric quadratic loss; symmetric linear loss; absolute mean scaled error; LnQ; ease of communication; ease of interpretation; ease of implementation; scale-independence; time series analysis; rolling-origin evaluation; inventory control; relative root mean squared error; RelRMSE; relative performance; forecast evaluation setup; data science; forecast density; mean-unbiasedness; median-unbiasedness; binomial test; Wilcoxon signed rank test; boxplots; double-scale plots (search for similar items in EconPapers)
Date: 2021-08-20
New Economics Papers: this item is included in nep-for
Note: View the original document on HAL open archive server: https://hal.science/hal-03359179
References: Add references at CitEc
Citations:
Published in International Journal of Statistics and Probability, 2021, 10 (5), pp.46-69. ⟨10.5539/ijsp.v10n5p46⟩
Downloads: (external link)
https://hal.science/hal-03359179/document (application/pdf)
Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.
Export reference: BibTeX
RIS (EndNote, ProCite, RefMan)
HTML/Text
Persistent link: https://EconPapers.repec.org/RePEc:hal:journl:hal-03359179
DOI: 10.5539/ijsp.v10n5p46
Access Statistics for this paper
More papers in Post-Print from HAL
Bibliographic data for series maintained by CCSD (hal@ccsd.cnrs.fr).