{"corpus_id":147703930,"paper_sha":"6f9dc6f8519e927d948a13aa7ae0df336f443eb9","doi":null,"arxiv_id":"1905.03222","pmid":null,"pmcid":null,"mag_id":2971169014,"dblp_id":"conf/nips/RomanoPC19","acl_id":null,"title":"Conformalized Quantile Regression","year":2019,"publication_date":"2019-05-08","venue":"Neural Information Processing Systems","journal":{"name":null,"pages":"3538-3548","volume":null},"journal_issn":null,"journal_title":null,"publication_types":["JournalArticle"],"pubmed_pub_types":null,"s2_fields_of_study":["Mathematics","Computer Science","Economics"],"reference_count":51,"citation_count":832,"influential_citation_count":159,"is_open_access":false,"arxiv_categories":["stat.ME","stat.ML"],"arxiv_license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","arxiv_journal_ref":null,"mesh_headings":null,"chemicals":null,"comments_corrections":null,"source_flags":1,"s2_open_access_pdf_url":null,"s2_open_access_landing_url":null,"s2_open_access_license":null,"s2_open_access_status":null,"pmc_open_access_pdf_url":null,"pmc_open_access_landing_url":null,"pmc_open_access_license":null,"pmc_open_access_status":null,"unpaywall_open_access_pdf_url":null,"unpaywall_open_access_landing_url":null,"unpaywall_open_access_license":null,"unpaywall_open_access_status":null,"abstract":"Conformal prediction is a technique for constructing prediction intervals that attain valid coverage in finite samples, without making distributional assumptions. Despite this appeal, existing conformal methods can be unnecessarily conservative because they form intervals of constant or weakly varying length across the input space. In this paper we propose a new method that is fully adaptive to heteroscedasticity. It combines conformal prediction with classical quantile regression, inheriting the advantages of both. We establish a theoretical guarantee of valid coverage, supplemented by extensive experiments on popular regression datasets. We compare the efficiency of conformalized quantile regression to other conformal methods, showing that our method tends to produce shorter intervals.","claims":[{"public_id":"cl_393fc9f58d384af3ffa3f680737a5cbe","status":"active","text":"A fully adaptive method to heteroscedasticity is proposed by combining conformal prediction with classical quantile regression.","confidence":0.97,"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/claims/cl_393fc9f58d384af3ffa3f680737a5cbe"},{"public_id":"cl_96564170023392db5b53d19076532b58","status":"active","text":"On popular regression datasets, conformalized quantile regression tends to produce shorter prediction intervals than other conformal methods.","confidence":0.95,"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/claims/cl_96564170023392db5b53d19076532b58"},{"public_id":"cl_be32be6ccab70345a060cfdbab97ffd5","status":"active","text":"Valid finite-sample coverage is theoretically guaranteed for the proposed conformalized quantile regression method.","confidence":0.96,"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/claims/cl_be32be6ccab70345a060cfdbab97ffd5"}],"concepts":[{"public_id":"co_598734e48ad8ce8710a119376ea22444","status":"active","name":"popular regression datasets","description":"Benchmark regression datasets used to evaluate predictive interval methods in experiments.","types":["evaluation dataset"],"aliases":[],"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/concepts/co_598734e48ad8ce8710a119376ea22444"},{"public_id":"co_6486cb30290ddc0b0fb51333a4d026e3","status":"active","name":"prediction intervals","description":"Intervals intended to contain future response values with a specified coverage level.","types":["output"],"aliases":[],"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/concepts/co_6486cb30290ddc0b0fb51333a4d026e3"},{"public_id":"co_7b563bedf0bf98a545a3819c3fa6634b","status":"active","name":"conformalized quantile regression","description":"A prediction-interval method that combines conformal prediction with quantile regression to adapt interval width across the input space.","types":["method"],"aliases":["CQR"],"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/concepts/co_7b563bedf0bf98a545a3819c3fa6634b"},{"public_id":"co_7e799084c21a7dfbe0c64de09be6bcf3","status":"active","name":"conformal methods","description":"Methods based on conformal prediction for producing calibrated uncertainty estimates or prediction intervals.","types":["method family"],"aliases":[],"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/concepts/co_7e799084c21a7dfbe0c64de09be6bcf3"},{"public_id":"co_88bd286b57de1f5d8a24fb2a2acbed91","status":"active","name":"conformal prediction","description":"A framework for constructing prediction sets or intervals with coverage guarantees from held-out calibration data.","types":["method"],"aliases":[],"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/concepts/co_88bd286b57de1f5d8a24fb2a2acbed91"},{"public_id":"co_983774c9e62584a316ff042c87b369fc","status":"active","name":"finite samples","description":"A setting in which guarantees are required to hold for a fixed, limited sample size.","types":["statistical setting"],"aliases":[],"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/concepts/co_983774c9e62584a316ff042c87b369fc"},{"public_id":"co_9c6ac5b91b3204b24427c076a539cccf","status":"active","name":"heteroscedasticity","description":"Input-dependent variability in the response noise or prediction uncertainty.","types":["statistical property"],"aliases":[],"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/concepts/co_9c6ac5b91b3204b24427c076a539cccf"},{"public_id":"co_ab718932637d81bcf966bf2f8d9d0192","status":"active","name":"valid coverage","description":"The property that prediction intervals contain the true response at the stated frequency.","types":["performance metric"],"aliases":[],"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/concepts/co_ab718932637d81bcf966bf2f8d9d0192"},{"public_id":"co_e1c6593203fc57d756687fcf08fcfa6f","status":"active","name":"quantile regression","description":"A regression technique that estimates conditional quantiles of the response given the inputs.","types":["method"],"aliases":[],"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/concepts/co_e1c6593203fc57d756687fcf08fcfa6f"}],"external_ids":{"DOI":null,"ArXiv":"1905.03222","PubMed":null,"PubMedCentral":null,"MAG":2971169014,"DBLP":"conf/nips/RomanoPC19","ACL":null},"open_access":{"is_open_access":true,"pdf_url":"https://arxiv.org/pdf/1905.03222","landing_url":"https://arxiv.org/abs/1905.03222","source":"arxiv","pdf_url_source":"derived_arxiv","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","reason":null},"reference_availability":{"status":"available","references_indexed":true,"full_text_available":true,"full_text_source":"arxiv","count_basis":"semantic_scholar_metadata","extraction_status":"not_applicable","reason":null},"source":{"provider":"episteme2","base_corpus":"semantic_scholar_dump","freshness_mode":"unknown","basis":["semantic_scholar_metadata","postgres_metadata"],"limits":["paper metadata is based on indexed upstream scholarly datasets","claims and concepts are available only for extracted papers","absence of claims or concepts means no extracted graph data is available in this response"],"status":"available","degraded":false,"degraded_reasons":[],"diagnostics":{"status":"available","degraded":false,"degraded_reasons":[],"metadata_status":"available","graph_status":"available","abstract_status":"available"},"source_flags":1},"paper_id":639050,"paper_uid":"d4de481e-8770-4faa-a880-5c08b3d9fa04","canonical_identity":{"paper_id":639050,"paper_uid":"d4de481e-8770-4faa-a880-5c08b3d9fa04","identity_status":"available","lookup_basis":"semantic_scholar_external_id","compatibility_path":"corpus_id"},"url":"https://sah.borca.ai/papers/147703930"}