Automatic identification of time series features for rule-based forecasting

Adya, Monica; Collopy, Fred; Armstrong, J. Scott; Kennedy, Miles

doi:10.1016/s0169-2070(01)00079-6

Cited by 81 publications

(68 citation statements)

References 11 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…This issue was addressed by Adya et al, (2001) who extended RBF by developing various automated routines to detect most time series features that were manually identified in C&As RBF. The design of this Automated RBF (ARBF) set the stage for further replication and validation of RBF on a much larger data set of 3003 time series in Adya et al, (2000) as well as an extension that involved a simpler set of forecasting rules (Adya & Lusk, 2013a).…”

Section: The Origins Of Cst: a History Of Replications And Extensionsmentioning

confidence: 99%

“…In the current world of streaming and big data, for a decision-maker to code time series features is manifestly impractical. This begs the next step -that is to integrate the elaborated CST system presented above into an expanded FDSS that captures the time series at its source/initial engagement and uses automate feature identification routines, such as those found in Adya et al, (2001). These time series characterizations can then feed into a forecasting system, such as C&As RBF that can select forecasting methods based on complexity as well as features of the time series.…”

Section: Implications For Fdss Designmentioning

confidence: 99%

See 1 more Smart Citation

Replication and Extension of a Forecasting Decision Support System: An Empirical Examination of the Time Series Complexity Scoring Technique

Adya¹,

Lusk²

2018

AIS TRR

Self Cite

View full text Add to dashboard Cite

Abstract:This study presents a conceptual replication of Adya and Lusk's (2016) forecasting decision support system (FDSS) that identifies the complexity or simplicity of a time series. Prior studies in forecasting have argued convincingly that the design of FDSS should incorporate the complexity of the forecasting task. Yet, there existed no formal way of determining time series complexity until this FDSS, referred to as the Complexity Scoring Technique (CST). The CST uses characteristics of the time series to trigger 12 rules that score the complexity of a time series and classify it along the binary dimension of Simple or Complex. The CST was originally validated using statistical forecasts of a small set of 54 time series as well as judgmental forecasts from 14 representative participants to confirm that the FDSS successfully distinguished Simple series from Complex ones. In this study, we (a) replicate the CST on a much larger set of data from both statistical and judgmental forecasting methods, and (b) extend and validate the series classification categories from the binary Simple-Complex used in the original CST to Very Simple, Simple, Complex, and Very Complex thus adding an ordinal link between the two previous binary designations. Findings suggest that both the replication and extension of the CST further validate it, thereby greatly enhancing its use in the practice of forecasting. Implications for research and practice are discussed.

show abstract

Section: The Origins Of Cst: a History Of Replications And Extensionsmentioning

confidence: 99%

Section: Implications For Fdss Designmentioning

confidence: 99%

Replication and Extension of a Forecasting Decision Support System: An Empirical Examination of the Time Series Complexity Scoring Technique

Adya¹,

Lusk²

2018

AIS TRR

Self Cite

View full text Add to dashboard Cite

show abstract

“…Their study presented the Rule-Based Forecasting (RBF) system, an FDSS that relies on 18 features of time series to combine forecasts from four accepted forecasting methods: Random Walk (Naïve 1), OLS Linear Regression, Holt's two parameter exponential smoothing (ARIMA [0,0,2]), and Brown's exponential smoothing. These initial set of C&A features were validated and extended in studies such as, 43,44,45 thereby establishing strong theoretical and empirical foundation over two decades.…”

Section: The Rule-based Forecasting Feature Setmentioning

confidence: 99%

“…Second, the feature codings were validated in several extensions. 43,44,45 Third, forecast errors for RBF and Combining A provided a priori validated benchmarks for refinement and sensitivity analysis of the CST during development. The assumption that complex series will have lower forecast accuracy than simple ones formed the logical basis for calibrations and directional hypothesis formation and testing.…”

Section: Overview Of Cst Development Processmentioning

confidence: 99%

See 1 more Smart Citation

Development and validation of a rule-based time series complexity scoring technique to support design of adaptive forecasting DSS

Adya

Lusk

2016

Decision Support Systems

Self Cite

View full text Add to dashboard Cite

Abstract:Evidence from forecasting research gives reason to believe that understanding time series complexity can enable design of adaptive forecasting decision support systems (FDSSs) to positively support forecasting behaviors and accuracy of outcomes. Yet, such FDSS design capabilities have not been formally explored because there exists no systematic approach to identifying series complexity. This study describes the NOT THE PUBLISHED VERSION; this is the author's final, peer-reviewed manuscript. The published version may be accessed by following the link in the citation at the bottom of the page.Decision Support Systems, Vol 83 (March 2016): pg. 70-82. DOI. This article is © Elsevier and permission has been granted for this version to appear in e-Publications@Marquette. Elsevier does not grant permission for this article to be further copied/distributed or hosted elsewhere without the express permission from Elsevier.2 development and validation of a rule-based complexity scoring technique (CST) that generates a complexity score for time series using 12 rules that rely on 14 features of series. The rule-based schema was developed on 74 series and validated on 52 holdback series using well-accepted forecasting methods as benchmarks. A supporting experimental validation was conducted with 14 participants who generated 336 structured judgmental forecasts for sets of series classified as simple or complex by the CST. Benchmark comparisons validated the CST by confirming, as hypothesized, that forecasting accuracy was lower for series scored by the technique as complex when compared to the accuracy of those scored as simple. The study concludes with a comprehensive framework for design of FDSS that can integrate the CST to adaptively support forecasters under varied conditions of series complexity. The framework is founded on the concepts of restrictiveness and guidance and offers specific recommendations on how these elements can be built in FDSS to support complexity.

show abstract

State of the Art

2021

Business Forecasting

View full text Add to dashboard Cite

Automatic identification of time series features for rule-based forecasting

Abstract: Rule-based forecasting (RBF) is

Cited by 81 publications

References 11 publications

Replication and Extension of a Forecasting Decision Support System: An Empirical Examination of the Time Series Complexity Scoring Technique

Replication and Extension of a Forecasting Decision Support System: An Empirical Examination of the Time Series Complexity Scoring Technique

Development and validation of a rule-based time series complexity scoring technique to support design of adaptive forecasting DSS

State of the Art

Contact Info

Product

Resources

About