Forecast Evaluation Techniques for I4.0 Systems

Davydenko, Andrey; Sai, Cuong; Shcherbakov, Maxim

doi:10.1007/978-3-030-66077-2_7

Cited by 4 publications

(4 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The result is a statistical framework (in the sense of the definition proposed in Davydenko and Charith, 2020) including the settings, criteria, methods and tools, and the workflow for the particular task of measuring forecast bias. Software implementation is straightforward and can be based on the flexible data formats proposed in (Davydenko et al, 2021). Almost any software environment (including Microsoft Excel) can be used to implement the simple methods proposed, but we recommend R because it allows flexible implementation of visual tools.…”

Section: Discussionmentioning

confidence: 99%

“…For example, we may wish to compare bias of forecasts obtained as a result of positive and negative judgmental adjustments to statistical forecasts, or to compare bias of forecasts obtained in different seasons or years. So, ideally, it should be possible to slice-and-dice forecast data and to obtain corresponding bias indicators for data subsets (see, e.g., Davydenko et al, 2021, for examples of constructing queries to subset forecast data). A procedure for conducting a formal statistical test for the presence of bias is also desirable.…”

Section: The Point Forecast Evaluation Setup (Pfes)mentioning

confidence: 99%

“…Further experiments (not discussed here for brevity) showed that the use of absolute median scaled errors may not give a reliable indicator of median bias. Another problem is that we may want to aggregate the AMScE across horizons and the arithmetic mean will lead to the over-influence of forecasts with greater horizons (Davydenko et al, 2021). There have been attempts to model regression bias based on the use of scaled errors and scaled forecasts (Fildes et al, 2009;Davydenko et al, 2010), but this approach may lead to spurious correlation due to the correlation between the error and the scale (Davydenko, 2012, pp.…”

Section: Scaled Errorsmentioning

confidence: 99%

“…12 above) to explore forecast data, see if data was loaded correctly, and detect potential data flaws. Use individual time series plots (such as the fixed origin or fixed horizon plots presented in Davydenko et al, 2021) to explore series with pronounced outliers.…”

Section: Forecast Evaluation Workflows (Fews)mentioning

confidence: 99%

See 3 more Smart Citations

Assessing Point Forecast Bias Across Multiple Time Series: Measures and Visual Tools

Davydenko¹,

Goodwin²

2021

Preprint

Self Cite

View full text Add to dashboard Cite

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. /// Note: This is the final version of the paper, which appeared in the International Journal of Statistics and Probability. The first draft of this paper was uploaded to Preprints.org on 11 May, 2021: https://doi.org/10.20944/preprints202105.0261.v1 /// Copyrights: This is an open access article distributed under the terms and conditions of the Creative Commons Attribution license (http://creativecommons.org/licenses/by/4.0/).

show abstract

Section: Discussionmentioning

confidence: 99%

Section: The Point Forecast Evaluation Setup (Pfes)mentioning

confidence: 99%