Adaptive Optimization for Prediction with Missing Data

When training predictive models on data with missing entries, the most widely used and versatile approach is a pipeline technique where we first impute missing entries and then compute predictions. In this paper, we view prediction with missing data as a two-stage adaptive optimization problem and p...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Bertsimas, Dimitris, Delarue, Arthur, Pauphilet, Jean
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext bestellen
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
container_end_page
container_issue
container_start_page
container_title
container_volume
creator Bertsimas, Dimitris
Delarue, Arthur
Pauphilet, Jean
description When training predictive models on data with missing entries, the most widely used and versatile approach is a pipeline technique where we first impute missing entries and then compute predictions. In this paper, we view prediction with missing data as a two-stage adaptive optimization problem and propose a new class of models, adaptive linear regression models, where the regression coefficients adapt to the set of observed features. We show that some adaptive linear regression models are equivalent to learning an imputation rule and a downstream linear regression model simultaneously instead of sequentially. We leverage this joint-impute-then-regress interpretation to generalize our framework to non-linear models. In settings where data is strongly not missing at random, our methods achieve a 2-10% improvement in out-of-sample accuracy.
doi_str_mv 10.48550/arxiv.2402.01543
format Article
fullrecord <record><control><sourceid>arxiv_GOX</sourceid><recordid>TN_cdi_arxiv_primary_2402_01543</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2402_01543</sourcerecordid><originalsourceid>FETCH-LOGICAL-a673-a45dcadf81470ddea96fdcc6cc8fdb868fa6463e80dbd5ed65a61e841fac3d553</originalsourceid><addsrcrecordid>eNotj7tuAjEQRd1QRMAHpMI_sIuN7VlTriAkkYigoF8NHhtG4iXvijy-Pskm1dFtrs4R4lGr0nrn1BTzB9_LmVWzUmlnzYOAmvDW8T3KzQ_O_IUdXy8yXbPc5kgc-vnO3VG-cdvy5SCX2OFIDBKe2jj-51DsVk-7xUux3jy_Lup1gVCZAq2jgJS8tpUiijiHRCFACD7R3oNPCBZM9Ir25CKBQ9DRW50wGHLODMXk77YXb26Zz5g_m9-Apg8w34aBQcM</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype></control><display><type>article</type><title>Adaptive Optimization for Prediction with Missing Data</title><source>arXiv.org</source><creator>Bertsimas, Dimitris ; Delarue, Arthur ; Pauphilet, Jean</creator><creatorcontrib>Bertsimas, Dimitris ; Delarue, Arthur ; Pauphilet, Jean</creatorcontrib><description>When training predictive models on data with missing entries, the most widely used and versatile approach is a pipeline technique where we first impute missing entries and then compute predictions. In this paper, we view prediction with missing data as a two-stage adaptive optimization problem and propose a new class of models, adaptive linear regression models, where the regression coefficients adapt to the set of observed features. We show that some adaptive linear regression models are equivalent to learning an imputation rule and a downstream linear regression model simultaneously instead of sequentially. We leverage this joint-impute-then-regress interpretation to generalize our framework to non-linear models. In settings where data is strongly not missing at random, our methods achieve a 2-10% improvement in out-of-sample accuracy.</description><identifier>DOI: 10.48550/arxiv.2402.01543</identifier><language>eng</language><subject>Computer Science - Learning ; Statistics - Machine Learning</subject><creationdate>2024-02</creationdate><rights>http://creativecommons.org/licenses/by-sa/4.0</rights><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>228,230,776,881</link.rule.ids><linktorsrc>$$Uhttps://arxiv.org/abs/2402.01543$$EView_record_in_Cornell_University$$FView_record_in_$$GCornell_University$$Hfree_for_read</linktorsrc><backlink>$$Uhttps://doi.org/10.48550/arXiv.2402.01543$$DView paper in arXiv$$Hfree_for_read</backlink></links><search><creatorcontrib>Bertsimas, Dimitris</creatorcontrib><creatorcontrib>Delarue, Arthur</creatorcontrib><creatorcontrib>Pauphilet, Jean</creatorcontrib><title>Adaptive Optimization for Prediction with Missing Data</title><description>When training predictive models on data with missing entries, the most widely used and versatile approach is a pipeline technique where we first impute missing entries and then compute predictions. In this paper, we view prediction with missing data as a two-stage adaptive optimization problem and propose a new class of models, adaptive linear regression models, where the regression coefficients adapt to the set of observed features. We show that some adaptive linear regression models are equivalent to learning an imputation rule and a downstream linear regression model simultaneously instead of sequentially. We leverage this joint-impute-then-regress interpretation to generalize our framework to non-linear models. In settings where data is strongly not missing at random, our methods achieve a 2-10% improvement in out-of-sample accuracy.</description><subject>Computer Science - Learning</subject><subject>Statistics - Machine Learning</subject><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2024</creationdate><recordtype>article</recordtype><sourceid>GOX</sourceid><recordid>eNotj7tuAjEQRd1QRMAHpMI_sIuN7VlTriAkkYigoF8NHhtG4iXvijy-Pskm1dFtrs4R4lGr0nrn1BTzB9_LmVWzUmlnzYOAmvDW8T3KzQ_O_IUdXy8yXbPc5kgc-vnO3VG-cdvy5SCX2OFIDBKe2jj-51DsVk-7xUux3jy_Lup1gVCZAq2jgJS8tpUiijiHRCFACD7R3oNPCBZM9Ir25CKBQ9DRW50wGHLODMXk77YXb26Zz5g_m9-Apg8w34aBQcM</recordid><startdate>20240202</startdate><enddate>20240202</enddate><creator>Bertsimas, Dimitris</creator><creator>Delarue, Arthur</creator><creator>Pauphilet, Jean</creator><scope>AKY</scope><scope>EPD</scope><scope>GOX</scope></search><sort><creationdate>20240202</creationdate><title>Adaptive Optimization for Prediction with Missing Data</title><author>Bertsimas, Dimitris ; Delarue, Arthur ; Pauphilet, Jean</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-a673-a45dcadf81470ddea96fdcc6cc8fdb868fa6463e80dbd5ed65a61e841fac3d553</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2024</creationdate><topic>Computer Science - Learning</topic><topic>Statistics - Machine Learning</topic><toplevel>online_resources</toplevel><creatorcontrib>Bertsimas, Dimitris</creatorcontrib><creatorcontrib>Delarue, Arthur</creatorcontrib><creatorcontrib>Pauphilet, Jean</creatorcontrib><collection>arXiv Computer Science</collection><collection>arXiv Statistics</collection><collection>arXiv.org</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext_linktorsrc</fulltext></delivery><addata><au>Bertsimas, Dimitris</au><au>Delarue, Arthur</au><au>Pauphilet, Jean</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Adaptive Optimization for Prediction with Missing Data</atitle><date>2024-02-02</date><risdate>2024</risdate><abstract>When training predictive models on data with missing entries, the most widely used and versatile approach is a pipeline technique where we first impute missing entries and then compute predictions. In this paper, we view prediction with missing data as a two-stage adaptive optimization problem and propose a new class of models, adaptive linear regression models, where the regression coefficients adapt to the set of observed features. We show that some adaptive linear regression models are equivalent to learning an imputation rule and a downstream linear regression model simultaneously instead of sequentially. We leverage this joint-impute-then-regress interpretation to generalize our framework to non-linear models. In settings where data is strongly not missing at random, our methods achieve a 2-10% improvement in out-of-sample accuracy.</abstract><doi>10.48550/arxiv.2402.01543</doi><oa>free_for_read</oa></addata></record>
fulltext fulltext_linktorsrc
identifier DOI: 10.48550/arxiv.2402.01543
ispartof
issn
language eng
recordid cdi_arxiv_primary_2402_01543
source arXiv.org
subjects Computer Science - Learning
Statistics - Machine Learning
title Adaptive Optimization for Prediction with Missing Data
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-02-19T00%3A50%3A59IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-arxiv_GOX&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=Adaptive%20Optimization%20for%20Prediction%20with%20Missing%20Data&rft.au=Bertsimas,%20Dimitris&rft.date=2024-02-02&rft_id=info:doi/10.48550/arxiv.2402.01543&rft_dat=%3Carxiv_GOX%3E2402_01543%3C/arxiv_GOX%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_id=info:pmid/&rfr_iscdi=true