FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W2970872277> ?p ?o ?g. }

• W2970872277 abstract "This thesis considers the problem of analogyidentification in the context of forecasting. We develop and test arange of segmentation approaches, with the aim of improving theaccuracy of forecasting methods that employ analogies. The firstmanuscript of the thesis outlines our core methodologicalframework. This framework describes a forecasting process thatintegrates a multicriteria segmentation approach using aweighted-sum method for the identification of analogies during thesegmentation stage. This combines the information from pastrealizations of a set of time series with information about thefactors that govern the patterns observed, at the level of thedistance function. Using simulated and real-world data, weillustrate that a concurrent consideration of multiple criteria atthe segmentation stage can help to achieve better clusteringresults, which feed forward to improved forecasting accuracy. Thispaper contributes to the first methodological framework for theforecasting of analogous time series. Mulcriterion segmentationapproaches demonstrate a significant improvement in the forecastingperformance compared to single-criterion segmentation methods. Thesecond manuscript focuses on discussing the model selection problemrelated to the use of multicriteria clustering approaches. Althoughmulticriteria approaches to clustering are advantageous to thefinal increase of forecasting accuracy, the use of these approachesintroduces the challenge of an additional model selection duringthe segmentation stage. This is because even for the same number ofclusters, multicriteria clustering approaches may return sets ofclustering solutions that reflect different trade-offs between theconflicting criteria. Therefore, this thesis also includes workaddressing the model selection problem for multicriteria clusteringin a forecasting context. We demonstrate that the quality ofclustering solutions is best assessed in the problem-specific(forecasting) context. Computationally, this is the most expensiveapproach, and we, therefore, describe a compromise, which uses astandard internal validation technique (the Silhouette Widthmeasure) for the determination of clusters, but performs weightselection based on the best average (historical) forecastingperformance of the forecasting algorithm. Further, the thirdmanuscript addresses instability issues stemming from theclustering procedures by integrating bagging techniques into theforecasting process. Segmentations of analogies have been reportedto give rise to further increase in the final forecasting accuracy,but the application of clustering techniques in the segmentationstage may result in instabilities related to the model selectionstep. By combining the forecasts derived from multiple models, theaggregated forecast is expected to lower down the uncertainty ofthe results via the aggregation scheme. We, therefore, employ thebootstrap aggregation techniques to further improve the forecastingprocess, and this results in a further boost to the…" @default.
• W2970872277 created "2019-09-05" @default.
• W2970872277 creator A5071966599 @default.
• W2970872277 date "2018-01-01" @default.
• W2970872277 modified "2023-09-27" @default.
• W2970872277 title "Segmentation approaches for the identification of analogies in a forecasting context" @default.
• W2970872277 hasPublicationYear "2018" @default.
• W2970872277 type Work @default.
• W2970872277 sameAs 2970872277 @default.
• W2970872277 citedByCount "0" @default.
• W2970872277 crossrefType "dissertation" @default.
• W2970872277 hasAuthorship W2970872277A5071966599 @default.
• W2970872277 hasConcept C116834253 @default.
• W2970872277 hasConcept C119857082 @default.
• W2970872277 hasConcept C124101348 @default.
• W2970872277 hasConcept C154945302 @default.
• W2970872277 hasConcept C166957645 @default.
• W2970872277 hasConcept C205649164 @default.
• W2970872277 hasConcept C2779343474 @default.
• W2970872277 hasConcept C41008148 @default.
• W2970872277 hasConcept C59822182 @default.
• W2970872277 hasConcept C73555534 @default.
• W2970872277 hasConcept C81917197 @default.
• W2970872277 hasConcept C86803240 @default.
• W2970872277 hasConcept C89600930 @default.
• W2970872277 hasConceptScore W2970872277C116834253 @default.
• W2970872277 hasConceptScore W2970872277C119857082 @default.
• W2970872277 hasConceptScore W2970872277C124101348 @default.
• W2970872277 hasConceptScore W2970872277C154945302 @default.
• W2970872277 hasConceptScore W2970872277C166957645 @default.
• W2970872277 hasConceptScore W2970872277C205649164 @default.
• W2970872277 hasConceptScore W2970872277C2779343474 @default.
• W2970872277 hasConceptScore W2970872277C41008148 @default.
• W2970872277 hasConceptScore W2970872277C59822182 @default.
• W2970872277 hasConceptScore W2970872277C73555534 @default.
• W2970872277 hasConceptScore W2970872277C81917197 @default.
• W2970872277 hasConceptScore W2970872277C86803240 @default.
• W2970872277 hasConceptScore W2970872277C89600930 @default.
• W2970872277 hasLocation W29708722771 @default.
• W2970872277 hasOpenAccess W2970872277 @default.
• W2970872277 hasPrimaryLocation W29708722771 @default.
• W2970872277 hasRelatedWork W122439171 @default.
• W2970872277 hasRelatedWork W1977936283 @default.
• W2970872277 hasRelatedWork W1992342484 @default.
• W2970872277 hasRelatedWork W2012292069 @default.
• W2970872277 hasRelatedWork W2087068564 @default.
• W2970872277 hasRelatedWork W2111392486 @default.
• W2970872277 hasRelatedWork W2121643442 @default.
• W2970872277 hasRelatedWork W2126718491 @default.
• W2970872277 hasRelatedWork W2152201775 @default.
• W2970872277 hasRelatedWork W2315008505 @default.
• W2970872277 hasRelatedWork W2395005906 @default.
• W2970872277 hasRelatedWork W245362991 @default.
• W2970872277 hasRelatedWork W2769177808 @default.
• W2970872277 hasRelatedWork W2792338380 @default.
• W2970872277 hasRelatedWork W2900724415 @default.
• W2970872277 hasRelatedWork W2921315464 @default.
• W2970872277 hasRelatedWork W2966810016 @default.
• W2970872277 hasRelatedWork W3156734410 @default.
• W2970872277 hasRelatedWork W3170253735 @default.
• W2970872277 hasRelatedWork W745719157 @default.
• W2970872277 isParatext "false" @default.
• W2970872277 isRetracted "false" @default.
• W2970872277 magId "2970872277" @default.
• W2970872277 workType "dissertation" @default.