MétaCan
Menu
Back to cohort
Record W4307948995 · doi:10.3390/forecast4040048

Precision and Reliability of Forecasts Performance Metrics

2022· article· en· W4307948995 on OpenAlex

Why this work is in the frame

A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.

affAt least one author lists a Canadian institution in the pinned OpenAlex snapshot.
fundA Canadian funder is recorded on the work.

Bibliographic record

VenueForecasting · 2022
Typearticle
Languageen
FieldDecision Sciences
TopicForecasting Techniques and Applications
Canadian institutionsUniversité de MontréalPolytechnique Montréal
FundersMitacs
KeywordsVariance (accounting)Reliability (semiconductor)Metric (unit)Computer scienceSensitivity (control systems)Noise (video)Series (stratigraphy)Selection (genetic algorithm)EconometricsQuality (philosophy)Model selectionPerformance metricStatisticsData miningReliability engineeringMachine learningArtificial intelligenceMathematicsEngineering

Abstract

fetched live from OpenAlex

The selection of an accurate performance metric is highly important to evaluate the quality of a forecasting method. This evaluation may help to select between different forecasting tools of forecasting outputs, and then support many decisions within a company. This paper proposes to evaluate the sensitivity and reliability of forecasts performance metrics. The methodology is tested using multiple time series of different scales and demand patterns, such as intermittent demand. The idea is to add to each series a noise following a known distribution to represent forecasting models of a known error distribution. Varying the parameters of the distribution of the noise allows to evaluate how sensitive and reliable performance metrics are to changes in bias and variance of the error of a forecasting model. The experiments concluded that sRMSE is more reliable than MASE in most cases on those series. sRMSE is especially reliable for detecting changes in the variance of a model and sPIS is the most sensitive metric to the bias of a model. sAPIS is sensible to both variance and bias but is less reliable.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.

Full frame distilled prediction

Teacher imitation

Not calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.

metaresearch head score (Codex)0.005
metaresearch head score (Gemma)0.004
Version: codex-gemma-dda1882f352aValidation status: machine_predicted_unvalidated
Candidate categoriesnone
Consensus categoriesnone
DomainCandidate signal: none · Consensus signal: none
Study designCandidate signal: Simulation or modeling · Consensus signal: none
GenreCandidate signal: Empirical · Consensus signal: Empirical
Teacher disagreement score0.882
Threshold uncertainty score0.491

Codex and Gemma teacher scores by category

CategoryCodexGemma
Metaresearch0.0050.004
Meta-epidemiology (narrow)0.0000.000
Meta-epidemiology (broad)0.0000.000
Bibliometrics0.0000.002
Science and technology studies0.0010.000
Scholarly communication0.0000.000
Open science0.0010.001
Research integrity0.0000.000
Insufficient payload (model declined to judge)0.0000.000

Machine scores (provisional)

The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.

Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.

Opus teacher head0.147
GPT teacher head0.357
Teacher spread0.210 · how far apart the two teachers sit on this one work
Validation statusscore_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it