[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-81835-en":3,"doc-seo-81835-105":31,"detail-sidebar-cat-0-en-105":93},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":20,"is_deleted":4,"is_public":21,"is_downloadable":21,"audit_status":21,"page_count":22,"language":23,"language_code":24,"site_id":25,"html_lang":24,"table_of_contents":26,"faqs":27,"seo_title":28,"seo_description":14,"update_tm":29,"read_time":30},81835,4398048950312,"Violet","https://ap-avatar.wpscdn.com/avatar/400002538284de19e3c?_k=1778320343897328908",8,"Research & Report","Predicting Heterogeneous Treatment Effects of Building Energy Saving Retrofits Using Causal Machine Learning","Machine learning models are increasingly used in information systems to predict outcomes in complex socio-technical settings, but predictive modeling does not automatically identify causal effects. This is critical for building retrofits, where energy-savings estimates must be unbiased for climate policy and investment. Because retrofit adoption correlates with characteristics that also drive energy consumption, naïve ML can produce biased effect estimates. The paper benchmarks causal ML estimators, constructs a physically grounded simulator with known truth, and finds DoubleML lowest errors, especially for complex envelope retrofits.","PREDICTING HETEROGENEOUS TREATMENT EFFECTS OF BUILDING ENERGY SAVING RETROFITS USING CAUSAL MACHINE LEARNING  \nCompleted Research Paper  \nKevin Zalipski, Paderborn University, Paderborn, Germany, [kevin.zalipski@upb.de](kevin.zalipski@upb.de)[ ](kevin.zalipski@upb.de)[David Zapata Gonzalez](David Zapata Gonzalez), Paderborn University, Paderborn, Germany, david.zapata@upb.de Oliver Müller, Paderborn University, Paderborn, Germany, [oliver.mueller@upb.de](oliver.mueller@upb.de)  \nAbstract  \nInformation Systems research increasingly relies on machine learning (ML) to predict outcomes in complex sociotechnical systems, yet predictive models are not designed to identify causal effects. This limitation is particularly critical in building retrofits, where unbiased estimates of energy savings are essential for climate policy and investment decisions. Because retrofit adoption is shaped by household and building characteristics that also affect energy consumption, predictive ML can yield biased effect estimates. This paper systematically benchmarks leading causal ML estimators, including metalearners (S-, T-and X-Learners) and DoubleML across multiple retrofit interventions. To enable this comparison, we construct a physically grounded simulation in which true treatment effects and realistic adoption biases are known. Results show that DoubleML achieves the lowest estimation errors, particularly for complex envelope retrofits. These findings demonstrate that orthogonalising the treatment assignment improves causal effect estimation and provides a methodological foundation for large-scale energy retrofit and policy evaluation.  \nKeywords: Causal Machine Learning, Double Machine Learning, Confounding, Heterogeneous Effects, Retrofitting.  \n1 Introduction  \nInformation Systems (IS) research has rapidly adopted machine learning (ML) methods to model socio-technical phenomena using large-scale, high-dimensional data, primarily to improve prediction performance (Abdel-Karim et al., 2021) . Yet many questions central to the IS discipline are inherently causal, for example: the effect of introducing a new system feature, changing a platform policy or deploying an IT-enabled intervention (Athey, 2017) . Standard predictive ML methods cannot answer such questions because they optimise for correlational accuracy rather than causal identification. This gap has motivated a rapidly expanding IS literature on causal ML, which combines the flexibility of ML with principled causal inference to estimate treatment effects and support causal reasoning in complex socio-technical settings (Kuzmanovic et al., 2024; Tafti & Shmueli, 2020, 2025; von Zahn et al., 2025; Zapata Gonzalez et al., 2025) . This limitation of standard predictive ML is particularly critical in energy informatics contexts such as building energy retrofits, where treatment assignments (e.g., which house receives the treatment) are highly biased.  \nBuildings account for a substantial share of global energy use and carbon emissions. According to the International Energy Agency (IEA) (2023), their operations represent around 30% of global final energy consumption and 26% of energy-related CO₂ emissions. In most countries, the residential sector is a major contributor to this impact (Eurostat, 2025; U.S. Energy Information Administration  \nThirty-Fourth European Conference on Information Systems (ECIS 2026), Milan, Italy 1  \nCausal Effects of Energy Retrofits  \n(EIA), 2024) . Energy retrofitting, through measures such as façade insulation, window replacement or heating-system upgrades, is therefore central to national and global decarbonisation strategies (Camarasa et al., 2022) . Accurate estimation of the energy savings achievable through such retrofitsis critical, since these estimates form the foundation for cost-benefit analysis, risk assessment and the prioritization of retrofit options.  \nTraditionally, retrofit evaluations rely on physical building-energy simulations, which model t","cbCaikklall011SA","https://ap.wps.com/l/cbCaikklall011SA","pdf",2345007,6,1,17,"English","en",105,"# Abstract\n# Introduction\n## Motivation: prediction vs causal questions\n## Why energy retrofit evaluation needs causality\n## Traditional simulation and its limitations\n## Role of causal machine learning","[{\"question\":\"Why can predictive machine learning produce biased retrofit effect estimates?\",\"answer\":\"Retrofit adoption is shaped by household and building characteristics that also influence energy consumption. When models optimize for correlation, confounding remains unaddressed, biasing estimated effects.\"},{\"question\":\"What causal ML approaches are benchmarked in the study?\",\"answer\":\"The paper benchmarks leading causal ML estimators, including metalearners such as S-, T-, and X-Learners and DoubleML, across multiple retrofit interventions.\"},{\"question\":\"How do the authors evaluate causal estimators in a controlled setting?\",\"answer\":\"They construct a physically grounded simulation where the true treatment effects and realistic adoption biases are explicitly modeled and therefore known, enabling direct benchmarking against ground truth.\"}]","Predicting Heterogeneous Treatment Effects of Building Energy Saving Retrofits Using Causal Machine Learning | PDF",1784176523,43,{"code":4,"msg":32,"data":33},"ok",{"site_id":25,"language":24,"slug":34,"title":13,"keywords":35,"description":14,"schema_data":36,"social_meta":88,"head_meta":90,"extra_data":92,"updated_unix":29},"predicting-heterogeneous-treatment-effects-of-building-energy-saving-retrofits-using-causal-machine-learning","",{"@graph":37,"@context":87},[38,55,70],{"@type":39,"itemListElement":40},"BreadcrumbList",[41,45,49,52],{"item":42,"name":43,"@type":44,"position":21},"https://docshare.wps.com","Home","ListItem",{"item":46,"name":47,"@type":44,"position":48},"https://docshare.wps.com/document/","Document",2,{"item":50,"name":12,"@type":44,"position":51},"https://docshare.wps.com/document/research-report/",3,{"item":53,"name":13,"@type":44,"position":54},"https://docshare.wps.com/document/predicting-heterogeneous-treatment-effects-of-building-energy-saving-retrofits-using-causal-machine-learning/81835/",4,{"url":53,"name":13,"@type":56,"author":57,"headline":13,"publisher":59,"fileFormat":62,"inLanguage":24,"description":14,"dateModified":63,"datePublished":64,"encodingFormat":62,"isAccessibleForFree":65,"interactionStatistic":66},"DigitalDocument",{"name":9,"@type":58},"Person",{"url":42,"name":60,"@type":61},"DocShare","Organization","application/pdf","2026-07-29","2026-07-16",true,{"@type":67,"interactionType":68,"userInteractionCount":20},"InteractionCounter",{"@type":69},"ViewAction",{"@type":71,"mainEntity":72},"FAQPage",[73,79,83],{"name":74,"@type":75,"acceptedAnswer":76},"Why can predictive machine learning produce biased retrofit effect estimates?","Question",{"text":77,"@type":78},"Retrofit adoption is shaped by household and building characteristics that also influence energy consumption. When models optimize for correlation, confounding remains unaddressed, biasing estimated effects.","Answer",{"name":80,"@type":75,"acceptedAnswer":81},"What causal ML approaches are benchmarked in the study?",{"text":82,"@type":78},"The paper benchmarks leading causal ML estimators, including metalearners such as S-, T-, and X-Learners and DoubleML, across multiple retrofit interventions.",{"name":84,"@type":75,"acceptedAnswer":85},"How do the authors evaluate causal estimators in a controlled setting?",{"text":86,"@type":78},"They construct a physically grounded simulation where the true treatment effects and realistic adoption biases are explicitly modeled and therefore known, enabling direct benchmarking against ground truth.","https://schema.org",{"og:url":53,"og:type":89,"og:title":13,"og:site_name":60,"og:description":14},"article",{"robots":91,"canonical":53},"index,follow",{"doc_id":7,"site_id":25},{"code":4,"msg":5,"data":94},[95,99,103,107,112,116,121,124,129,132,136],{"id":21,"doc_module":4,"doc_module_name":47,"category_name":96,"show_sort_weight":97,"slug":98},"Story & Novel",90,"story-novel",{"id":48,"doc_module":4,"doc_module_name":47,"category_name":100,"show_sort_weight":101,"slug":102},"Literature",80,"literature",{"id":54,"doc_module":4,"doc_module_name":47,"category_name":104,"show_sort_weight":105,"slug":106},"Exam",70,"exam",{"id":108,"doc_module":4,"doc_module_name":47,"category_name":109,"show_sort_weight":110,"slug":111},5,"Comic",60,"comic",{"id":20,"doc_module":4,"doc_module_name":47,"category_name":113,"show_sort_weight":114,"slug":115},"Technology",50,"technology",{"id":117,"doc_module":4,"doc_module_name":47,"category_name":118,"show_sort_weight":119,"slug":120},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":47,"category_name":12,"show_sort_weight":122,"slug":123},30,"research-report",{"id":125,"doc_module":4,"doc_module_name":47,"category_name":126,"show_sort_weight":127,"slug":128},9,"Religion & Spirituality",20,"religion-spirituality",{"id":127,"doc_module":4,"doc_module_name":47,"category_name":130,"show_sort_weight":127,"slug":131},"World Cup","world-cup",{"id":133,"doc_module":4,"doc_module_name":47,"category_name":134,"show_sort_weight":133,"slug":135},10,"Lifestyle","lifestyle",{"id":137,"doc_module":4,"doc_module_name":47,"category_name":138,"show_sort_weight":108,"slug":139},19,"General","general"]