A Study on Enhancing Hierarchical Time Series Forecasting employing Machine Learning Models

A Study on Enhancing Hierarchical Time Series Forecasting employing Machine Learning Models | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article A Study on Enhancing Hierarchical Time Series Forecasting employing Machine Learning Models Rudhir Chandra Mahalik, Sibarama Panigrahi This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-4991584/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Hierarchical forecasting (HF) methods are extensively utilized for precise decision-making by providing coherent forecasts across various levels. Traditionally, statistical models have been employed in HF. However, these static approaches often overlook the dynamic nature of the series during the aggregation and disaggregation of aperiodic and spontaneous components. This paper addresses this issue by leveraging the dynamic and nonlinear modeling capabilities of machine learning (ML) models in HF. Specifically, we implement and evaluate the performance of seventeen ML models at each hierarchical level, reconciling them post-forecasting using top-down (TD), bottom-up (BU), middle-out (MO), min trace (MT), and optimal combination (OC) approaches for one-step-ahead and seven different direct multi-step-ahead forecasting of the M5 competition dataset. Extensive non-parametric statistical analyses are conducted to rank the ML models for HF and address ten research questions pertaining to HF. Simulation results suggest that the k-nearest neighbors regression (KNNR) model and BU approach provide statistically superior performance across all pairs of ML model and HF approach considering one to eight-step-ahead forecasting. It is also observed that employing ML models at specific hierarchical levels, followed by reconciliation statistically improves the forecasting accuracy at all levels of the hierarchy. Physical sciences/Engineering Physical sciences/Engineering/Electrical and electronic engineering Hierarchical Time Series Forecasting Machine Learning M5 Competition Reconcilation Full Text Additional Declarations No competing interests reported. Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-4991584","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":359008091,"identity":"2084a6aa-229c-44d7-8634-4a0ea9285640","order_by":0,"name":"Rudhir Chandra Mahalik","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA90lEQVRIie3RsUrEQBDG8VkOkma8tHNE1ldYSGPnq2Q5iE2w3kouHGx1b2DhK1xpYTFHCpt9AO1MczYKpvM650SsZFPYWOy/DPtjPghAKvUPy/JuBWBIF/kKGKA+/fpcR8gceyHuvFpsGJihxkmi6VJIcHb7KM9+SHQY2G48eKrg6W3Xj/cNQt4/w3AXI7t1eeJJq5urmnnfImBjwIYIUZ2fKbkyK1vDzE6GtQDWR4i8VzLM+kX4JsXrBMmUBwxkN4RHIsNo6gqqdYmOKkIZFrjBjPaGY+Ts9mUYP8y1vngI1bvjpS6K5TAcIuSXu3D8P6lUKpX6U5+YuVCxICmH3wAAAABJRU5ErkJggg==","orcid":"","institution":"National Institute of Technology Rourkela","correspondingAuthor":true,"prefix":"","firstName":"Rudhir","middleName":"Chandra","lastName":"Mahalik","suffix":""},{"id":359008092,"identity":"bab5ab32-919a-4c92-b413-3401c6915756","order_by":1,"name":"Sibarama Panigrahi","email":"","orcid":"","institution":"National Institute of Technology Rourkela","correspondingAuthor":false,"prefix":"","firstName":"Sibarama","middleName":"","lastName":"Panigrahi","suffix":""}],"badges":[],"createdAt":"2024-08-28 13:53:23","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-4991584/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-4991584/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":73348112,"identity":"e0cf8ca6-3a2d-4c79-9250-b759e4f74879","added_by":"auto","created_at":"2025-01-09 06:47:40","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2057668,"visible":true,"origin":"","legend":"","description":"","filename":"Manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-4991584/v1_covered_06f89dc9-5a1a-420d-b12d-85a0f718e389.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"A Study on Enhancing Hierarchical Time Series Forecasting employing Machine Learning Models","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":true,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":true,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Hierarchical Time Series Forecasting, Machine Learning, M5 Competition, Reconcilation","lastPublishedDoi":"10.21203/rs.3.rs-4991584/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-4991584/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eHierarchical forecasting (HF) methods are extensively utilized for precise decision-making by providing coherent forecasts across various levels. Traditionally, statistical models have been employed in HF. However, these static approaches often overlook the dynamic nature of the series during the aggregation and disaggregation of aperiodic and spontaneous components. This paper addresses this issue by leveraging the dynamic and nonlinear modeling capabilities of machine learning (ML) models in HF. Specifically, we implement and evaluate the performance of seventeen ML models at each hierarchical level, reconciling them post-forecasting using top-down (TD), bottom-up (BU), middle-out (MO), min trace (MT), and optimal combination (OC) approaches for one-step-ahead and seven different direct multi-step-ahead forecasting of the M5 competition dataset. Extensive non-parametric statistical analyses are conducted to rank the ML models for HF and address ten research questions pertaining to HF. Simulation results suggest that the k-nearest neighbors regression (KNNR) model and BU approach provide statistically superior performance across all pairs of ML model and HF approach considering one to eight-step-ahead forecasting. It is also observed that employing ML models at specific hierarchical levels, followed by reconciliation statistically improves the forecasting accuracy at all levels of the hierarchy.\u003c/p\u003e","manuscriptTitle":"A Study on Enhancing Hierarchical Time Series Forecasting employing Machine Learning Models","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2024-10-08 09:54:36","doi":"10.21203/rs.3.rs-4991584/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"02997d5e-996b-4ddd-93bf-8629dc1c2641","owner":[],"postedDate":"October 8th, 2024","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":38199158,"name":"Physical sciences/Engineering"},{"id":38199159,"name":"Physical sciences/Engineering/Electrical and electronic engineering"}],"tags":[],"updatedAt":"2025-01-09T06:39:19+00:00","versionOfRecord":[],"versionCreatedAt":"2024-10-08 09:54:36","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-4991584","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-4991584","identity":"rs-4991584","version":["v1"]},"buildId":"qtupq5eGEP_6zYnWcrvyt","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}