[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"detail-sidebar-cat-0-en-105":3,"doc-seo-290831-105":59,"doc-detail-290831-en":130},{"code":4,"msg":5,"data":6},0,"success",[7,13,18,23,28,33,38,43,48,51,55],{"id":8,"doc_module":4,"doc_module_name":9,"category_name":10,"show_sort_weight":11,"slug":12},1,"Document","Story & Novel",90,"story-novel",{"id":14,"doc_module":4,"doc_module_name":9,"category_name":15,"show_sort_weight":16,"slug":17},2,"Literature",80,"literature",{"id":19,"doc_module":4,"doc_module_name":9,"category_name":20,"show_sort_weight":21,"slug":22},4,"Exam",70,"exam",{"id":24,"doc_module":4,"doc_module_name":9,"category_name":25,"show_sort_weight":26,"slug":27},5,"Comic",60,"comic",{"id":29,"doc_module":4,"doc_module_name":9,"category_name":30,"show_sort_weight":31,"slug":32},6,"Technology",50,"technology",{"id":34,"doc_module":4,"doc_module_name":9,"category_name":35,"show_sort_weight":36,"slug":37},7,"Healthcare",40,"healthcare",{"id":39,"doc_module":4,"doc_module_name":9,"category_name":40,"show_sort_weight":41,"slug":42},8,"Research & Report",30,"research-report",{"id":44,"doc_module":4,"doc_module_name":9,"category_name":45,"show_sort_weight":46,"slug":47},9,"Religion & Spirituality",20,"religion-spirituality",{"id":46,"doc_module":4,"doc_module_name":9,"category_name":49,"show_sort_weight":46,"slug":50},"World Cup","world-cup",{"id":52,"doc_module":4,"doc_module_name":9,"category_name":53,"show_sort_weight":52,"slug":54},10,"Lifestyle","lifestyle",{"id":56,"doc_module":4,"doc_module_name":9,"category_name":57,"show_sort_weight":24,"slug":58},19,"General","general",{"code":4,"msg":60,"data":61},"ok",{"site_id":62,"language":63,"slug":64,"title":65,"keywords":66,"description":67,"schema_data":68,"social_meta":123,"head_meta":125,"extra_data":127,"updated_unix":129},105,"en","explainability-of-machine-learning-models-under-missing-data","Explainability of Machine Learning Models under Missing Data","","Missing data is a prevalent issue that can significantly impair model performance and explainability. This paper summarizes the development of missing-data research in the context of Explainable Artificial Intelligence and experimentally studies how different imputation methods affect SHAP (Shapley Additive exPlanations). It compares strategies by feature importance and interaction and provides a theoretical analysis of missing values’ impact on Shapley values, showing that imputation choice can bias explanations. The study also finds that lower test prediction MSE does not guarantee lower MSE in Shapley values, and that applying XGBoost directly to missing data can harm explainability compared with imputing first, offering practical guidance for selecting techniques.",{"@graph":69,"@context":122},[70,84,105],{"@type":71,"itemListElement":72},"BreadcrumbList",[73,77,79,82],{"item":74,"name":75,"@type":76,"position":8},"https://docshare.wps.com","Home","ListItem",{"item":78,"name":9,"@type":76,"position":14},"https://docshare.wps.com/document/",{"item":80,"name":30,"@type":76,"position":81},"https://docshare.wps.com/document/technology/",3,{"item":83,"name":65,"@type":76,"position":19},"https://docshare.wps.com/document/explainability-of-machine-learning-models-under-missing-data/290831/",{"url":83,"name":65,"@type":85,"image":86,"author":91,"headline":65,"publisher":94,"fileFormat":97,"inLanguage":63,"description":67,"dateModified":98,"datePublished":99,"encodingFormat":97,"isAccessibleForFree":100,"interactionStatistic":101},"DigitalDocument",{"url":87,"@type":88,"width":89,"height":90},"https://docshare.wps.com/thumbnails/explainability-of-machine-learning-models-under-missing-data/290831.png","ImageObject",300,407,{"name":92,"@type":93},"Skyler","Person",{"url":74,"name":95,"@type":96},"DocShare","Organization","application/pdf","2026-10-14","2026-09-17",true,{"@type":102,"interactionType":103,"userInteractionCount":29},"InteractionCounter",{"@type":104},"ViewAction",{"@type":106,"mainEntity":107},"FAQPage",[108,114,118],{"name":109,"@type":110,"acceptedAnswer":111},"Why is missing data a critical problem for explainable AI?","Question",{"text":112,"@type":113},"Missing data can impair both predictive performance and the reliability of explanations. Imputation choices can introduce biases or distortions that affect interpretability results.","Answer",{"name":115,"@type":110,"acceptedAnswer":116},"How does this work evaluate imputation methods for explanations?",{"text":117,"@type":113},"The paper experimentally studies how different imputation strategies change SHAP-based feature importance and interaction, and it also provides a theoretical analysis of how missing values affect Shapley values.",{"name":119,"@type":110,"acceptedAnswer":120},"Does better predictive MSE always mean better SHAP explanations?",{"text":121,"@type":113},"No. The study shows that a lower test prediction MSE does not necessarily correspond to a lower MSE in Shapley values, and vice versa.","https://schema.org",{"og:url":83,"og:type":124,"og:title":65,"og:site_name":95,"og:description":67},"article",{"robots":126,"canonical":83},"index,follow",{"doc_id":128,"site_id":62},290831,1790176902,{"code":4,"msg":5,"data":131},{"doc_id":128,"user_id":132,"nickname":92,"user_avatar":133,"doc_module":4,"category_id":29,"category_name":30,"doc_title":65,"doc_description":67,"doc_content":134,"file_id":135,"file_url":136,"file_type":137,"file_size":138,"view_count":29,"is_deleted":4,"is_public":8,"is_downloadable":8,"audit_status":8,"page_count":139,"language":140,"language_code":63,"site_id":62,"html_lang":63,"table_of_contents":141,"faqs":142,"seo_title":143,"seo_description":67,"update_tm":144,"read_time":145},2336464648746,"https://ap-avatar.wpscdn.com/davatar_276721f389ce27ea32af1340a28f341c","arXiv :2407 .00411v3 [ cs .LG] 22 Jan 2025  \nExplainability of Machine Learning Models under  \nMissing Data  \nTuan L. Vo∗c, Thu Nguyen∗1a, Luis M. Lopez-Ramosa , Hugo L. Hammerb,a , Michael A. Rieglera,b , P˚al Halvorsena,b  \naSimulaMet, Oslo, Norway  \nb Oslo Metropolitan University, Oslo, Norway c LTCI, T´el´ecom Paris, Institut Polytechnique de Paris, Paris, France  \nAbstract  \nMissing data is a prevalent issue that can significantly impair model performance and explainability. This paper briefly summarizes the development of the field of missing data with respect to Explainable Artificial Intelligence and experimentally investigates the effects of various imputation methods on SHAP (SHapley Additive exPlanations), a popular technique for explaining the output of complex machine learning models. Next, we compare different imputation strategies and assess their impact on feature importance and interaction as determined by Shapley values. Moreover, we also theoretically analyze the effects of missing values on Shapley values. Importantly, our findings reveal that the choice of imputation method can introduce biases that could lead to changes in the Shapley values, thereby affecting the explainability of the model. Moreover, we also show that a lower test prediction MSE (Mean Square Error) does not necessarily imply a lower MSE in Shapley values and vice versa. Also, while XGBoost (eXtreme Gradient Boosting) is a method that could handle missing data directly, using XGBoost directly on missing data can seriously affect explainability compared to imputing the data before training XGBoost. This study provides a comprehensive evaluation of imputation methods in the context of model explanations, offering practical guidance for selecting appropriate techniques based on dataset characteristics and analysis objectives. The results underscore the importance of considering imputation effects to ensure robust and reliable insights from machine learning models.  \n1∗denotes equal contribution  \nPreprint submitted to Arxiv January 23, 2025  \nKeywords: Missing Data, Imputation, Model Explainability  \n1. Introduction  \nMissing data is a common issue that can significantly affect model performance and explainability. This problem can occur due to various reasons, such as data collection errors, privacy concerns, or intentional omission. Addressing missing data is crucial, and one effective approach is imputation, i.e., filling in the missing data points to create a more complete dataset for analysis, thereby improving the overall reliability of the results.  \nVarious imputation methods exist, ranging from simple techniques like mean imputation to more sophisticated approaches such as multiple imputation by chained equations (MICE) and k-nearest neighbors (KNN) . Each method has its strengths and weaknesses, influencing not only the performance of predictive models but also their explainability. For example, in [1], the authors studied the effects of missing data on the correlation heatmap and found out that the technique that computes a correlation matrix with the highest accuracy (in terms of root mean square error, RMSE) does not always produce correlation plots that closely resemble those derived from complete data. However, the effects of missing data on various aspects of explainable AI (XAI) have not yet been fully studied.  \nIn terms of explainability, Shapley values, a concept derived from cooperative game theory, have gained prominence as a robust method for interpreting complex models [2] . Shapley values attribute the contribution of each feature to the final prediction, offering insights into feature importance and interaction. However, the accuracy and reliability of Shapley values can be affected by the choice of the imputation method, as the imputed values can introduce biases or distortions. Despite the importance of this issue, little attention has been paid to the effect of imputation on the Shapley values of the downstream ","cbCaid920ZBzabe3","https://ap.wps.com/l/cbCaid920ZBzabe3","pdf",5746845,49,"English","# Introduction\n## Problem and motivation: missing data impacts performance and explainability\n## Imputation methods and their influence on explainable AI\n## Shapley values for model interpretation and risks from imputation bias\n## Paper contributions and organization\n# Related works and methods\n## Review of XAI and missing data\n## Methods examined","[{\"question\":\"Why is missing data a critical problem for explainable AI?\",\"answer\":\"Missing data can impair both predictive performance and the reliability of explanations. Imputation choices can introduce biases or distortions that affect interpretability results.\"},{\"question\":\"How does this work evaluate imputation methods for explanations?\",\"answer\":\"The paper experimentally studies how different imputation strategies change SHAP-based feature importance and interaction, and it also provides a theoretical analysis of how missing values affect Shapley values.\"},{\"question\":\"Does better predictive MSE always mean better SHAP explanations?\",\"answer\":\"No. The study shows that a lower test prediction MSE does not necessarily correspond to a lower MSE in Shapley values, and vice versa.\"}]","Explainability of Machine Learning Models under Missing Data | PDF",1789642695,123]